{"schema_version":"1.5.1","types":{"text":"A string.","integer":"A whole number.","number":"A decimal number, as the source states it. We never round it.","timestamp":"An ISO-8601 timestamp with its zone (Z or ±hh:mm). Stamps we write are UTC. Stamps read from a vendor keep the vendor’s offset.","date":"A calendar date, YYYY-MM-DD.","json":"Structured data. Nested in the JSON export and JSON text in a CSV cell, except where the meaning says the column is JSON text in both.","boolean":"true or false, as the words true and false in a CSV cell."},"provenance":"Every row ends with source_url, the URL it was read from, and fetched_at, when that fetch ran. source_runs has started_at instead, since a failed run fetched nothing. entity_securities has checked_on, since its rows are hand-checked. entity_identifiers also gives each identifier its own source URL and fetch time. In a CSV an empty cell is null.","license":"Rows dated in the last 90 days are CC BY 4.0. A table that grows over time downloads only those rows by default. Add ?history=full for every row. That download adds a license column before source_url: CC-BY-NC-4.0 for a row dated before the window that we first published on or after 2026-11-04, CC-BY-4.0 for the rest. Tables of current rows and lookups download whole. Rebuilt at an as_of before the window, they get the same column. Upstream data keeps its own license.","time":"Event time is when the thing happened, as the source dates it. Knowledge time is when the row first existed for us (or, for a filing, when it was filed). ?as_of= cuts on knowledge time.","joining":"Every provider and company column holds an entity_id from entity_identifiers. That table has each company’s CIK, LEI, ticker, MIC, and FIGI, each with the URL and time we read it. A lab without listed shares of its own (OpenAI, Google, xAI) reaches a security through entity_securities, which names the parent or disclosed investor and the filing that says so. Tables with a cik or ticker column join on that too. A provider of other is a company we don’t identify.","tables":[{"name":"snapshots","description":"One row per fetch of every source, with its content hash, HTTP status, byte count, and the R2 key of the raw response.","csv":"/export/snapshots.csv","json":"/export/snapshots.json","key":["id"],"grain":"One row per fetch of a source.","cadence":"Model catalogs up to every 5 minutes, SEC feeds every 30, everything else every 6 hours.","time":{"event":null,"knowledge":"fetched_at"},"as_of":{"mode":"cut","note":"Rows with fetched_at at or before as_of."},"history":{"mode":"windowed","dated":["fetched_at"],"full":"/export/snapshots.csv?history=full","note":"Downloads the last 90 days by default, dated by fetched_at, when we read the page. Add ?history=full for every row.","start":"From our first check of each source."},"joins":[],"licensing":{"commercial":"yes","note":"Yes. These are records of our own checks. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":[],"upstream":[],"not_cleared":[]},"columns":[{"name":"id","type":"text","meaning":"The fetch: <source_id>:<fetched_at>.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the fetched body.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"raw_key","type":"text","meaning":"Key of the raw body in this site’s object store (R2). A fetch that found the same content points at the existing object.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"http_status","type":"integer","meaning":"The response’s HTTP status. Null where none was recorded.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"bytes","type":"integer","meaning":"Size of the fetched body. Null where none was recorded.","unit":"bytes","nullable":true,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL fetched.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch ran, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prices_latest","description":"The current price and catalog row per SKU and source, as each vendor page stated it at our last check.","csv":"/export/prices_latest.csv","json":"/export/prices_latest.json","key":["entity_key","source_id"],"grain":"One row per SKU per source, as of the last check.","cadence":"Model catalogs every 5 minutes when they change, every source every 6 hours.","time":{"event":"effective_from","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"whole","dated":[],"full":null,"note":"Downloads whole, since it holds one current row per SKU.","start":"Current rows only. Past prices are in changes (from our first check) and archived_prices (Internet Archive copies from January 2023)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 4 are ours to license and 14 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-pricing-html","google-models-md","mistral-models-md","mistral-pricing-md"],"upstream":[],"not_cleared":["openai-models-md","openai-pricing-md","anthropic-models-md","anthropic-pricing-md","xai-models-md","meta-models-md","meta-muse-wonder-md","meta-muse-spark-md","meta-pricing-md","deepseek-pricing-md","alibaba-pricing-html","alibaba-models-html","moonshot-models-md","openrouter-models"]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, model:<provider>:<model_slug>, with :<tier> when the SKU has one. With source_id it identifies the row. A models page and a pricing page can both list a SKU, so one entity_key can have two rows.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"model_slug","type":"text","meaning":"The model’s identifier as the vendor’s page prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"tier","type":"text","meaning":"The price tier, where a model has several (Google’s standard, batch, flex, and priority, or xAI’s long_context). Null for a single-tier SKU.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"context_window","type":"text","meaning":"The context window as printed (“500k”, “1M”), not converted to tokens. Null where the page states none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"input_per_mtok","type":"number","meaning":"List price of input tokens. Null where the page prints none.","unit":"USD per million tokens","nullable":true,"role":null,"as_of_only":false},{"name":"cached_input_per_mtok","type":"number","meaning":"List price of cached input tokens (a cache hit). Null where the page prints none.","unit":"USD per million tokens","nullable":true,"role":null,"as_of_only":false},{"name":"output_per_mtok","type":"number","meaning":"List price of output tokens. Null where the page prints none.","unit":"USD per million tokens","nullable":true,"role":null,"as_of_only":false},{"name":"currency","type":"text","meaning":"Always USD.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"effective_from","type":"date","meaning":"The day a scheduled price takes effect, where the page announces one. Null for the price in force.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"effective_until","type":"date","meaning":"The day the page says the price in force is replaced. Null when no change is scheduled.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"notes","type":"text","meaning":"Qualifiers the page prints beside the row (cache-write prices, promotions, deprecations), verbatim. Null where it prints none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"jobs_open","description":"Every role currently listed on the job boards we track. A closed role leaves this table and shows up in changes.","csv":"/export/jobs_open.csv","json":"/export/jobs_open.json","key":["entity_key","source_id"],"grain":"One row per open role per job board.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"published_at","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"whole","dated":[],"full":null,"note":"Downloads whole, since it holds the roles listed now.","start":"Current roles only. Roles that closed are in changes, from our first check. Open-role counts before then are in archived_hiring (Internet Archive copies from January 2023)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["openai-ashby","anthropic-greenhouse","anthropic-greenhouse-departments","xai-greenhouse","xai-greenhouse-departments","mistral-ashby","google-careers"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, job:<provider>:<job_id>. With source_id it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"job_id","type":"text","meaning":"The board’s own job id: an Ashby UUID, a Greenhouse number, or a Google careers id.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"title","type":"text","meaning":"The role title, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"department","type":"text","meaning":"The department as the board names it (on Google’s feed, its categories joined by commas). Null where the board gives none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"team","type":"text","meaning":"The team within the department. Ashby boards only. Null elsewhere.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"location","type":"text","meaning":"The location text, several locations in one string, verbatim. Null where the board gives none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"employment_type","type":"text","meaning":"The employment type as the board labels it (FullTime, …). Null where the board does not carry one, as Greenhouse does not.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"published_at","type":"timestamp","meaning":"When the board says the role was first published, with the board’s own offset.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"updated_at","type":"timestamp","meaning":"When the board says the listing last changed, with its own offset. Null where the board gives none (Google’s feed).","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"company_name","type":"text","meaning":"The employer the board prints, verbatim. Every xAI row says SpaceXAI: that board lists SpaceX and xAI roles together.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"incidents","description":"Every status-page incident we’ve stored, with the vendor’s title, impact, and status verbatim, start and resolution, and components. Impact is null where the feed publishes none (Anthropic’s Atom feed doesn’t). We keep an incident after it drops out of its feed.","csv":"/export/incidents.csv","json":"/export/incidents.json","key":["entity_key","source_id"],"grain":"One row per status-page incident.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"started_at","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"windowed","dated":["started_at","first_seen_at"],"full":"/export/incidents.csv?history=full","note":"Downloads the last 90 days by default, dated by started_at, when the incident began (first_seen_at where the feed gives no start). Add ?history=full for every row.","start":"What each feed listed at our first check, and every incident since."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 1 is ours to license and 6 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-cloud-status"],"upstream":[],"not_cleared":["openai-status","anthropic-status","mistral-status","deepseek-status","deepseek-status-atom","xai-status"]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, incident:<provider>:<incident_id>. With source_id it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"incident_id","type":"text","meaning":"The status page’s own incident id.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"title","type":"text","meaning":"The incident title, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"impact","type":"text","meaning":"The vendor’s impact or severity word, verbatim and not mapped across vendors. Statuspage uses none, minor, major, and critical. Google Cloud uses low, medium, and high. Null where the feed publishes none, as Anthropic’s Atom feed does.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"status","type":"text","meaning":"The vendor’s status word, verbatim. Statuspage uses investigating, identified, monitoring, resolved, and postmortem. Google Cloud gives the latest update’s status (AVAILABLE once resolved).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"started_at","type":"timestamp","meaning":"When the incident started, as the feed states it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"resolved_at","type":"timestamp","meaning":"When the incident was resolved, as the feed states it. Null while open.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"components","type":"json","meaning":"The affected components (Statuspage) or products (Google Cloud): an array of names, sorted, possibly empty.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"incident_url","type":"text","meaning":"The incident’s own page on the status site.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"disclosures","description":"Every entry on a lab’s own misalignment-disclosure index (only OpenAI publishes one). Each row has the report or notice, title, summary, model, and context verbatim, and the date the index prints beside it. A count measures what a lab disclosed, not how its models behaved.","csv":"/export/disclosures.csv","json":"/export/disclosures.json","key":["entity_key","source_id"],"grain":"One row per entry on a lab’s misalignment-disclosure index.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"listed_on","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"windowed","dated":["listed_on","first_seen_at"],"full":"/export/disclosures.csv?history=full","note":"Downloads the last 90 days by default, dated by listed_on, the date the index prints (first_seen_at where it prints none). Add ?history=full for every row.","start":"Every entry the index lists, back to its first."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"no","note":"No. It stays on the free site and isn’t in any paid offer.","ours":[],"upstream":[],"not_cleared":["openai-misalignment-reports"]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, disclosure:<provider>:<kind>:<entry_id>. With source_id it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"kind","type":"text","meaning":"report or notice.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entry_id","type":"text","meaning":"The entry’s id on the index: a report’s slug, or a notice’s id without “notice-”.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"title","type":"text","meaning":"The entry’s title, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"summary","type":"text","meaning":"The index’s one-sentence summary, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"model","type":"text","meaning":"The model the report concerns, verbatim. Null on a notice.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"observed_during","type":"text","meaning":"The context the behavior was observed in, verbatim. Null on a notice.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"listed_on","type":"date","meaning":"The date the index prints beside the entry: a report’s last update, a notice’s posting date. It moves when a report is revised.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entry_url","type":"text","meaning":"The entry’s own page.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"disclosure_timelines","description":"The dates printed on each report’s own page. That’s every “Label: value” line verbatim, the discovery date, and the disclosure date where the page prints one.","csv":"/export/disclosure_timelines.csv","json":"/export/disclosure_timelines.json","key":["entity_key","source_id"],"grain":"One row per disclosure report page.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"disclosed_on","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"windowed","dated":["disclosed_on","discovered_on","first_seen_at"],"full":"/export/disclosure_timelines.csv?history=full","note":"Downloads the last 90 days by default, dated by disclosed_on, else discovered_on, else first_seen_at. Add ?history=full for every row.","start":"Every report the index links to."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"no","note":"No. It stays on the free site and isn’t in any paid offer.","ours":[],"upstream":[],"not_cleared":["openai-misalignment-reports"]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, disclosure_timeline:<provider>:<entry_id>. With source_id it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entry_id","type":"text","meaning":"The report’s slug. It joins disclosures.entry_id.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"title","type":"text","meaning":"The report page’s own title, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"lines","type":"json","meaning":"Every “Label: value” line of the page’s metadata, verbatim and in page order, minus “Report updated”: an array of {label, value} objects.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"discovered_on","type":"date","meaning":"The discovery date, from the line under any label the vendor uses for it, when that value is a single date. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"disclosed_on","type":"date","meaning":"A printed disclosure date, when the page has one that is a single date.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"filers","description":"One row per lab whose CIK we found in its own registration filing (sources.yaml pending_filers). Each row has the CIK, EDGAR’s name for the filer, and the filing we read it from.","csv":"/export/filers.csv","json":"/export/filers.json","key":["entity_key","source_id"],"grain":"One row per lab whose CIK we found in its own registration filing.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"resolved_file_date","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"whole","dated":[],"full":null,"note":"Downloads whole, since it’s a lookup of one CIK per lab.","start":"From the first registration filing we found for each lab."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, filer:<provider>, one per lab. With source_id it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"cik","type":"text","meaning":"The filer’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"name","type":"text","meaning":"EDGAR’s display name for the filer, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"resolved_form","type":"text","meaning":"The form of the filing the CIK was read from (S-1, S-1/A).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"resolved_accession","type":"text","meaning":"That filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"resolved_file_date","type":"date","meaning":"The day that filing was filed.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"revenue_facts","description":"One row per XBRL revenue fact from EDGAR company facts, with the filer, tag, period, value in USD, and the filing that reported it. Rows are never edited. A restatement is a new row under its own accession.","csv":"/export/revenue_facts.csv","json":"/export/revenue_facts.json","key":["entity_key","source_id"],"grain":"One row per XBRL revenue fact per filing.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"end","knowledge":"first_seen_at"},"as_of":{"mode":"rebuilt","note":"Rebuilt from the current rows and the change log. For each key we take the content the first change after as_of replaced, else the current row if it was written by then. The view gains an as_of_verified column. False marks a key whose state the log does not confirm."},"history":{"mode":"windowed","dated":["end"],"full":"/export/revenue_facts.csv?history=full","note":"Downloads the last 90 days by default, dated by end, the period end. Add ?history=full for every row.","start":"Every revenue fact in EDGAR company facts, as far back as each filer tagged it."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-companyfacts-spcx"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The row’s key across polls, revenue:<cik>:<tag>:<accession_no>:<start>:<end>. With source_id it identifies the row. A segment fact’s key ends with its segment member (:goog:GoogleCloudMember). There’s no segment column, so the key is what tells a segment value from the consolidated one.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"cik","type":"text","meaning":"The filer’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The filer’s name as EDGAR or the filing gives it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"taxonomy","type":"text","meaning":"us-gaap for an XBRL fact (the SEC’s tagged financial data). prospectus for a figure read off the statement of operations in a registration filing that tags no financials.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"tag","type":"text","meaning":"The XBRL revenue tag, or for prospectus the statement’s row label verbatim. Filers switch tags between years, so a series can span several.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"unit","type":"text","meaning":"Always USD.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"start","type":"date","meaning":"First day of the period the value covers.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"end","type":"date","meaning":"Last day of the period the value covers.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"val","type":"number","meaning":"The filer’s own value for the period, in US dollars, never scaled or rounded.","unit":"USD","nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The filing that reported the value. A restatement is a new row under its own accession.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"fy","type":"integer","meaning":"The fiscal year as the filer tagged it.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"fp","type":"text","meaning":"The fiscal period as the filer tagged it: FY or Q1 to Q4.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The reporting filing’s form type (10-K, 10-Q, S-1, …).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the reporting filing was filed.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"frame","type":"text","meaning":"SEC’s calendar frame (CY2026Q2) when this fact is the canonical one for it. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"whitelist_cik","type":"text","meaning":"The CIK again when the filer is on the sources.yaml whitelist of filers whose filings alert. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the row’s content. It changes when the content does, and each change is a row in changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id of the fetch whose parse produced the row’s current content.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of_verified","type":"boolean","meaning":"With ?as_of= only. False where the newest change at or before as_of disagrees with the rebuilt row, so the row may not be what the store held then.","unit":null,"nullable":false,"role":null,"as_of_only":true},{"name":"source_url","type":"text","meaning":"The URL the row’s current content was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prospectus_statements","description":"Every line of the statement of operations in a lab’s own registration filing (S-1, S-1/A). One row per filing, period, and line, with the line verbatim, a standard key where we have one, and the value in USD, shares, or USD per share. Null means the filing printed a dash. Interim periods are unaudited.","csv":"/export/prospectus_statements.csv","json":"/export/prospectus_statements.json","key":["entity_key","source_id","part"],"grain":"One row per filing, period, and statement line.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"filed","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. This view is built from the current filing rows, and we can’t rebuild its past versions yet."},"history":{"mode":"windowed","dated":["filed"],"full":"/export/prospectus_statements.csv?history=full","note":"Downloads the last 90 days by default, dated by filed, the filing date. Add ?history=full for every row.","start":"From each lab’s first registration filing we read (Anthropic’s S-1 of 2026-10-26)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The filing’s entity, statement:<cik>:<accession_no>:<start>:<end>. With source_id and part it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab that filed the registration statement.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"part","type":"integer","meaning":"The row’s position within its entity, from 0, in the order the filing prints it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"cik","type":"text","meaning":"The registrant’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The registrant’s name as the filing prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The registration filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form type: S-1 or S-1/A.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the filing was filed with the SEC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"start","type":"date","meaning":"First day of the period.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"end","type":"date","meaning":"Last day of the period.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"unaudited","type":"boolean","meaning":"True for an interim period, which registration statements print unaudited.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"units","type":"text","meaning":"The statement’s units line, verbatim (“in millions, except per share data”). val is already scaled out of it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"line_key","type":"text","meaning":"A canonical key where the label is a known one (revenue, total_operating_expenses, operating_income, net_income, …), else the label slugged.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"label","type":"text","meaning":"The line’s label as the statement prints it, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"val","type":"number","meaning":"The line’s value in whole units, scaled out of the printed millions or thousands. Negative where the statement prints parentheses, null where it prints a dash.","unit":"USD, USD per share, or shares, as unit says","nullable":true,"role":null,"as_of_only":false},{"name":"unit","type":"text","meaning":"USD, USD/share, or shares.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The registration document on EDGAR the row was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prospectus_commitments","description":"Contractual commitments printed in a lab’s registration filing. That covers the contractual-obligations table in MD&A (management’s discussion and analysis), the purchase-commitment schedule in the commitments note, and the one-figure totals, as of each date stated. One row per schedule line and year bucket, with bucket “Total” for the line’s total. approximate marks a figure the filing calls approximate, and statement quotes its sentence.","csv":"/export/prospectus_commitments.csv","json":"/export/prospectus_commitments.json","key":["entity_key","source_id","part"],"grain":"One row per filing, schedule line, and year bucket.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"filed","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. This view is built from the current filing rows, and we can’t rebuild its past versions yet."},"history":{"mode":"windowed","dated":["filed"],"full":"/export/prospectus_commitments.csv?history=full","note":"Downloads the last 90 days by default, dated by filed, the filing date. Add ?history=full for every row.","start":"From each lab’s first registration filing we read (Anthropic’s S-1 of 2026-10-26)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The filing’s entity, commitment:<accession_no>:<schedule>:<as_of, or undated>. With source_id and part it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab that filed the registration statement.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"part","type":"integer","meaning":"The row’s position within its entity, from 0, in the order the filing prints it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"cik","type":"text","meaning":"The registrant’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The registrant’s name as the filing prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The registration filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form type: S-1 or S-1/A.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the filing was filed with the SEC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"schedule","type":"text","meaning":"contractual_obligations (MD&A’s table), purchase_commitments (the commitments note’s audited schedule), purchase_commitments_interim (the note’s one-sentence interim total), or obligations_aggregate (MD&A’s one-sentence total).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of","type":"date","meaning":"The date the schedule is stated as of. Null when its sentence names none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"audited","type":"boolean","meaning":"True for a schedule in the audited notes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"approximate","type":"boolean","meaning":"True when the filing itself calls the figure approximate.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"line","type":"text","meaning":"The obligation’s label, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"bucket","type":"text","meaning":"When the amount falls due, as the filing labels it (“2026”, “Thereafter”), or Total for the line’s total.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"val","type":"number","meaning":"The amount in US dollars, scaled out of the printed millions or thousands.","unit":"USD","nullable":false,"role":null,"as_of_only":false},{"name":"statement","type":"text","meaning":"For a one-figure schedule, the sentence it was read from, verbatim. Null for a table.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The registration document on EDGAR the row was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prospectus_concentration","description":"Customers at or above 10% of revenue or of accounts receivable, from a lab’s registration filing. One row per filing, basis, customer, and period or balance date, with customers named as the filing names them. A null pct is the filing’s “*”, which means under 10% (not zero).","csv":"/export/prospectus_concentration.csv","json":"/export/prospectus_concentration.json","key":["entity_key","source_id","part"],"grain":"One row per filing, basis, customer, and period.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"filed","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. This view is built from the current filing rows, and we can’t rebuild its past versions yet."},"history":{"mode":"windowed","dated":["filed"],"full":"/export/prospectus_concentration.csv?history=full","note":"Downloads the last 90 days by default, dated by filed, the filing date. Add ?history=full for every row.","start":"From each lab’s first registration filing we read (Anthropic’s S-1 of 2026-10-26)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The filing’s entity, concentration:<accession_no>:<basis>. With source_id and part it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab that filed the registration statement.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"part","type":"integer","meaning":"The row’s position within its entity, from 0, in the order the filing prints it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"cik","type":"text","meaning":"The registrant’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The registrant’s name as the filing prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The registration filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form type: S-1 or S-1/A.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the filing was filed with the SEC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"basis","type":"text","meaning":"revenue or accounts_receivable.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"customer","type":"text","meaning":"The customer as the filing names it (“Customer A”).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"start","type":"date","meaning":"First day of the revenue period. Null for a receivables balance.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"end","type":"date","meaning":"Last day of the revenue period, or the balance date.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"unaudited","type":"boolean","meaning":"True for an unaudited interim column.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"pct","type":"number","meaning":"The customer’s share of revenue or of receivables. Null is the filing’s “*”: under 10%. It isn’t zero.","unit":"percent, 0 to 100","nullable":true,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The registration document on EDGAR the row was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prospectus_shares","description":"Share counts from a lab’s registration filing. Each row of “The Offering” (shares offered, and each class outstanding after the offering, with the underwriters’ option where stated), plus the pre-offering counts it’s based on. Null while the filing leaves a count blank.","csv":"/export/prospectus_shares.csv","json":"/export/prospectus_shares.json","key":["entity_key","source_id","part"],"grain":"One row per filing and share-count line.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"filed","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. This view is built from the current filing rows, and we can’t rebuild its past versions yet."},"history":{"mode":"windowed","dated":["filed"],"full":"/export/prospectus_shares.csv?history=full","note":"Downloads the last 90 days by default, dated by filed, the filing date. Add ?history=full for every row.","start":"From each lab’s first registration filing we read (Anthropic’s S-1 of 2026-10-26)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The filing’s entity, shares:<accession_no>. With source_id and part it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab that filed the registration statement.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"part","type":"integer","meaning":"The row’s position within its entity, from 0, in the order the filing prints it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"cik","type":"text","meaning":"The registrant’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The registrant’s name as the filing prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The registration filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form type: S-1 or S-1/A.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the filing was filed with the SEC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"measure","type":"text","meaning":"the_offering for a row of “The Offering” summary. before_offering for a pre-offering count it is based on.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"as_of","type":"date","meaning":"The date a before_offering count is stated as of. Null on the_offering rows.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"label","type":"text","meaning":"The row’s label, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"shares","type":"integer","meaning":"The count in whole shares. Null while the filing leaves it blank.","unit":"shares","nullable":true,"role":null,"as_of_only":false},{"name":"with_option","type":"integer","meaning":"The count if the underwriters exercise their option in full, where the row states one. Null on before_offering rows.","unit":"shares","nullable":true,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The registration document on EDGAR the row was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"prospectus_risk_factors","description":"Every risk-factor heading in a lab’s registration filing, verbatim and in order. Each is compared with the same filer’s previous filing and marked unchanged, added, reworded (previous holds the earlier words), or removed. A removed heading is one the previous filing printed and this one doesn’t, listed after this filing’s own.","csv":"/export/prospectus_risk_factors.csv","json":"/export/prospectus_risk_factors.json","key":["entity_key","source_id","part"],"grain":"One row per filing and risk-factor heading.","cadence":"Every 30 minutes with the SEC feeds, and again every 6 hours.","time":{"event":"filed","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. This view is built from the current filing rows, and we can’t rebuild its past versions yet."},"history":{"mode":"windowed","dated":["filed"],"full":"/export/prospectus_risk_factors.csv?history=full","note":"Downloads the last 90 days by default, dated by filed, the filing date. Add ?history=full for every row.","start":"From each lab’s first registration filing we read (Anthropic’s S-1 of 2026-10-26)."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba"],"upstream":[],"not_cleared":[]},"columns":[{"name":"entity_key","type":"text","meaning":"The filing’s entity, risk_factors:<accession_no>. With source_id and part it identifies the row.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"provider","type":"text","meaning":"The lab that filed the registration statement.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"part","type":"integer","meaning":"The row’s position within its entity, from 0, in the order the filing prints it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"cik","type":"text","meaning":"The registrant’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_name","type":"text","meaning":"The registrant’s name as the filing prints it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The registration filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form type: S-1 or S-1/A.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The day the filing was filed with the SEC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"section","type":"text","meaning":"The “Risks Related to …” subsection the heading sits under, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"heading","type":"text","meaning":"The risk-factor heading, verbatim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"change","type":"text","meaning":"Against the same filer’s previous filing: unchanged, added, reworded, or removed (a heading only the previous filing printed). Null on a filer’s first filing.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"previous","type":"text","meaning":"For a reworded heading, the previous filing’s words, verbatim. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When a fetch first produced the row, in UTC. It never changes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The registration document on EDGAR the row was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the fetch that produced the row’s current content ran, in UTC. A later fetch that finds the same content doesn’t move it. source_runs has every check.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"changes","description":"The change log. Every tracked item added, removed, or modified, with its before and after values. Rows are never edited. OpenRouter ranking rows are left out (see rankings_daily on /data).","csv":"/export/changes.csv","json":"/export/changes.json","key":["id"],"grain":"One row per tracked item added, removed, or modified.","cadence":"A check that finds a difference adds rows. Rows are never edited.","time":{"event":null,"knowledge":"detected_at"},"as_of":{"mode":"cut","note":"Rows with detected_at at or before as_of."},"history":{"mode":"windowed","dated":["detected_at"],"full":"/export/changes.csv?history=full","note":"Downloads the last 90 days by default, dated by detected_at, when we saw the change. Add ?history=full for every row.","start":"From our first check of each source."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 23 are ours to license, 1 passes on under upstream terms (CC BY 4.0), and 21 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-pricing-html","google-models-md","mistral-models-md","mistral-pricing-md","openai-ashby","anthropic-greenhouse","anthropic-greenhouse-departments","xai-greenhouse","xai-greenhouse-departments","mistral-ashby","google-careers","edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba","edgar-companyfacts-spcx","google-cloud-status"],"upstream":[{"id":"openrouter-rankings-daily","license":"CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)"}],"not_cleared":["openai-models-md","openai-pricing-md","anthropic-models-md","anthropic-pricing-md","xai-models-md","meta-models-md","meta-muse-wonder-md","meta-muse-spark-md","meta-pricing-md","deepseek-pricing-md","alibaba-pricing-html","alibaba-models-html","moonshot-models-md","openrouter-models","openai-status","anthropic-status","mistral-status","deepseek-status","deepseek-status-atom","xai-status","openai-misalignment-reports"]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of the entity key and the two content hashes, so the same change is never logged twice. A filed_fact change hashes its series, period, and accession instead.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"entity_key","type":"text","meaning":"The changed row’s entity_key, as the views above use it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_type","type":"text","meaning":"The kind of row: model, job, incident, disclosure, disclosure_timeline, filer, revenue, filing, document, offering, statement, commitment, concentration, shares, risk_factors, recommendation, or filed_fact (a hyperscaler series point a new filing moved, cited by its alert). OpenRouter ranking rows are left out.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"change_type","type":"text","meaning":"added, removed, or modified.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"before_hash","type":"text","meaning":"The row’s content_hash before the change. Null when added.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"after_hash","type":"text","meaning":"The row’s content_hash after the change. Null when removed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"before_json","type":"json","meaning":"JSON text (a string in the JSON export too): the row’s content before the change. Null when added.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"after_json","type":"json","meaning":"JSON text (a string in the JSON export too): the row’s content after the change. Null when removed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"detected_at","type":"timestamp","meaning":"When the fetch that observed the change ran, in UTC. Same instant as fetched_at.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL whose fetch observed the change.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that fetch ran, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"archived_prices","description":"Price history from before we watched the labs: one row per SKU per Internet Archive copy of the lab’s own pricing page that we could read. source_url is the copy, captured_at when the Archive saw the page, fetched_at when we read the copy. Never mixed into the live tables.","csv":"/export/archived_prices.csv","json":"/export/archived_prices.json","key":["id"],"grain":"One row per SKU per Internet Archive copy we could read.","cadence":"Once per backfill run, not on a schedule.","time":{"event":"captured_at","knowledge":"fetched_at"},"as_of":{"mode":"cut","note":"Rows with captured_at at or before as_of: when the Internet Archive saw the page, not when we read the copy (fetched_at, always later). So a backtest gets the page as it stood then, which this site did not hold at the time."},"history":{"mode":"windowed","dated":["captured_at"],"full":"/export/archived_prices.csv?history=full","note":"Downloads the last 90 days by default, dated by captured_at, when the Internet Archive saw the page. Add ?history=full for every row.","start":"Archived only. From the oldest copy we could read. The backfill looks at one copy a week from January 2023 to 2026-08-25."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"no","note":"No. It stays on the free site and isn’t in any paid offer.","ours":[],"upstream":[],"not_cleared":["wayback-pricing"]},"columns":[{"name":"id","type":"text","meaning":"The first 32 hex characters of the SHA-256 of capture_id and entity_key.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"capture_id","type":"text","meaning":"The copy the row was read from: <era>:<14-digit Wayback timestamp>, as in archive_captures.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_key","type":"text","meaning":"The SKU: model:<provider>:<slug>[:<tier>], the same key the live prices use, so /prices/<key> shows this row before the live ones.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of payload. Two copies with the same hash printed the same row.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"payload","type":"json","meaning":"The row as the page printed it, in the prices_latest fields: model_slug, tier, the three $/MTok prices (null where the page printed none), currency, and notes.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"captured_at","type":"timestamp","meaning":"When the Internet Archive captured the page, in UTC: the moment these prices were on it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"original_url","type":"text","meaning":"The lab’s own pricing page the Archive captured.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"basis","type":"text","meaning":"Always archived: read from a copy, not observed by this site.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The Wayback Machine copy (https://web.archive.org/web/<timestamp>id_/<original_url>).","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When we read the copy back from the Archive, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"archived_hiring","description":"Hiring history from before we read the labs’ job boards: open roles per Internet Archive copy of each lab’s own board that we could read. One board row per copy, and one row per department when the copy lists every role. source_url is the copy, captured_at when the Archive saw the board, fetched_at when we read the copy. Never mixed into the live tables.","csv":"/export/archived_hiring.csv","json":"/export/archived_hiring.json","key":["id"],"grain":"One board row per Internet Archive copy of a lab’s job board we could read, and one row per department when the copy lists every role.","cadence":"Once per backfill run, not on a schedule.","time":{"event":"captured_at","knowledge":"fetched_at"},"as_of":{"mode":"cut","note":"Rows with captured_at at or before as_of: when the Internet Archive saw the board, not when we read the copy (fetched_at, always later). So a backtest gets the board as it stood then, which this site did not hold at the time."},"history":{"mode":"windowed","dated":["captured_at"],"full":"/export/archived_hiring.csv?history=full","note":"Downloads the last 90 days by default, dated by captured_at, when the Internet Archive saw the board. Add ?history=full for every row.","start":"Archived only. From the oldest copy we could read (OpenAI 2023-01-08). The backfill looks at one copy a week from January 2023 to the day we began reading each board."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"no","note":"No. It stays on the free site and isn’t in any paid offer.","ours":[],"upstream":[],"not_cleared":["wayback-hiring"]},"columns":[{"name":"id","type":"text","meaning":"The first 32 hex characters of the SHA-256 of capture_id and the row’s scope or department.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"capture_id","type":"text","meaning":"The copy the count was read from: <era>:<14-digit Wayback timestamp>, as in archive_captures.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"scope","type":"text","meaning":"board (open roles on the whole board; one per copy) or department (one per department, when the copy lists every role; they sum to the board row).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"label","type":"text","meaning":"The department as the board printed it then. Null on the board row.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"department_id","type":"text","meaning":"The board’s own id for the department (Greenhouse, Ashby), where the copy prints one. Null on the board row and on Lever boards.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"department","type":"text","meaning":"The live department the row files under: today’s label for the same department_id where the board’s id ties them, else the label as printed. Null on the board row.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"open_roles","type":"integer","meaning":"Open roles the copy listed (department rows) or stated (board row).","unit":"roles","nullable":false,"role":null,"as_of_only":false},{"name":"captured_at","type":"timestamp","meaning":"When the Internet Archive captured the board, in UTC: the moment these roles were open on it.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"original_url","type":"text","meaning":"The lab’s own job board the Archive captured.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"basis","type":"text","meaning":"Always archived: read from a copy, not observed by this site.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The Wayback Machine copy (https://web.archive.org/web/<timestamp>id_/<original_url>).","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When we read the copy back from the Archive, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"archive_captures","description":"Every Internet Archive copy the price- and hiring-history backfill considered: read (with how many rows it gave) or skipped, with the reason. A skipped copy gave no rows; nothing on it was guessed.","csv":"/export/archive_captures.csv","json":"/export/archive_captures.json","key":["id"],"grain":"One row per Internet Archive copy the backfill considered.","cadence":"Once per backfill run, not on a schedule.","time":{"event":"captured_at","knowledge":"fetched_at"},"as_of":{"mode":"cut","note":"Rows with captured_at at or before as_of: when the Internet Archive saw the page, not when we read the copy."},"history":{"mode":"windowed","dated":["captured_at"],"full":"/export/archive_captures.csv?history=full","note":"Downloads the last 90 days by default, dated by captured_at, when the Internet Archive saw the page. Add ?history=full for every row.","start":"Archived only. One copy a week per lab pricing page or job board, from January 2023 to the day we began reading it."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"no","note":"No. It stays on the free site and isn’t in any paid offer.","ours":[],"upstream":[],"not_cleared":["wayback-pricing","wayback-hiring"]},"columns":[{"name":"id","type":"text","meaning":"The copy: <era>:<14-digit Wayback timestamp>.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"era","type":"text","meaning":"Which shape of the lab’s pricing page or job board the copy is read as (layers/terminal/server/pipeline/archive/eras.ts, hiring-eras.ts).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"provider","type":"text","meaning":"The lab the row belongs to: openai, anthropic, google, xai, meta, mistral, deepseek, alibaba, moonshot, or other.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"original_url","type":"text","meaning":"The lab’s own pricing page or job board the Archive captured.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"captured_at","type":"timestamp","meaning":"When the Internet Archive captured the page, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"status","type":"text","meaning":"parsed (its rows are in archived_prices or archived_hiring, by axis) or skipped (no rows; detail says why).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"detail","type":"text","meaning":"Why a skipped copy was skipped, in the parser’s or the fetch’s own words. On a read job board, what it held beyond the counts (a paged list: total only).","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"content_hash","type":"text","meaning":"SHA-256 (hex) of the copy’s text. Null when the fetch failed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"raw_key","type":"text","meaning":"Key of the saved copy in this site’s object store (R2). Null when the fetch failed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"bytes","type":"integer","meaning":"Size of the copy as read. Null when the fetch failed.","unit":"bytes","nullable":true,"role":null,"as_of_only":false},{"name":"rows","type":"integer","meaning":"How many rows the copy gave, in archived_prices or archived_hiring (see axis).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"axis","type":"text","meaning":"pricing (a lab’s pricing page; rows in archived_prices) or hiring (a lab’s job board; rows in archived_hiring).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The Wayback Machine copy (https://web.archive.org/web/<timestamp>id_/<original_url>).","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When we read the copy (or tried to), in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"alerts","description":"Every alert fired, with the change ids it cites and the rule that produced it.","csv":"/export/alerts.csv","json":"/export/alerts.json","key":["id"],"grain":"One row per alert.","cadence":"Written when a check finds a change that meets a rule. Judged alerts follow when the judge runs.","time":{"event":null,"knowledge":"created_at"},"as_of":{"mode":"cut","note":"Alerts with created_at at or before as_of. A judge alert (rule agent-judge) is written by a judge run after the changes it cites, and is held back until a judge run that accepted alerts had started (judge_runs.started_at). One that a later run wrote about an older change can still appear early, because an alert row doesn’t record which run wrote it."},"history":{"mode":"windowed","dated":["created_at"],"full":"/export/alerts.csv?history=full","note":"Downloads the last 90 days by default, dated by created_at, when the alert was written. Add ?history=full for every row.","start":"From our first check of each source."},"joins":[],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 23 are ours to license, 1 passes on under upstream terms (CC BY 4.0), and 21 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-pricing-html","google-models-md","mistral-models-md","mistral-pricing-md","openai-ashby","anthropic-greenhouse","anthropic-greenhouse-departments","xai-greenhouse","xai-greenhouse-departments","mistral-ashby","google-careers","edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba","edgar-companyfacts-spcx","google-cloud-status"],"upstream":[{"id":"openrouter-rankings-daily","license":"CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)"}],"not_cleared":["openai-models-md","openai-pricing-md","anthropic-models-md","anthropic-pricing-md","xai-models-md","meta-models-md","meta-muse-wonder-md","meta-muse-spark-md","meta-pricing-md","deepseek-pricing-md","alibaba-pricing-html","alibaba-models-html","moonshot-models-md","openrouter-models","openai-status","anthropic-status","mistral-status","deepseek-status","deepseek-status-atom","xai-status","openai-misalignment-reports"]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of what the alert is about. Its page is /alerts/<id>.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"severity","type":"text","meaning":"info, notable, or critical.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"headline","type":"text","meaning":"The alert in one line.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"explanation","type":"text","meaning":"Why it matters. It cites only fields in the changes it cites.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"change_ids","type":"json","meaning":"JSON text (a string in the JSON export too): an array of the changes.id it cites.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"rule","type":"text","meaning":"What wrote it: agent-judge (the judge routine), or a deterministic rule: s1-floor, periodic-floor, cik-resolved, model-floor, prospectus-floor, hiring-rollup, filed-fact-floor.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"created_at","type":"timestamp","meaning":"When the alert counts from: the detection of the newest change it cites, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL of the change the alert is stamped from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that change’s fetch ran, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"claim_readings","description":"One row per thesis claim and data date: the measured value its kill condition is checked against, with the detail behind it (docs/theses/).","csv":"/export/claim_readings.csv","json":"/export/claim_readings.json","key":["claim_id","date"],"grain":"One row per thesis claim and data date.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"date","knowledge":"computed_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. A claim reading is rewritten in place while its data is revised, and the value it replaced is not kept."},"history":{"mode":"windowed","dated":["date"],"full":"/export/claim_readings.csv?history=full","note":"Downloads the last 90 days by default, dated by date, the data date. Add ?history=full for every row.","start":"From the first data date each claim’s recipe reads."},"joins":[],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 42 are ours to license, 20 pass on under upstream terms (CC BY 4.0), and 35 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-pricing-html","google-models-md","mistral-models-md","mistral-pricing-md","openai-ashby","anthropic-greenhouse","anthropic-greenhouse-departments","xai-greenhouse","xai-greenhouse-departments","mistral-ashby","google-careers","edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba","edgar-companyfacts-spcx","google-cloud-status","sec-revenue-net","sec-revenue-akam","sec-revenue-fsly","greenhouse-cloudflare","gpu-lambda","gpu-coreweave","gpu-nebius","gpu-crusoe","gpu-together","gpu-digitalocean","gpu-verda","gpu-azure","sec-revenue-crwv","sec-msft-openai","sec-amzn-anthropic","pay-openai","pay-anthropic","newsroom-google-deepmind","newsroom-mistral"],"upstream":[{"id":"openrouter-rankings-daily","license":"CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)"},{"id":"npm-wrangler","license":null},{"id":"npm-openai","license":null},{"id":"npm-anthropic-sdk","license":null},{"id":"npm-google-genai","license":null},{"id":"npm-mistral","license":null},{"id":"npm-claude-code","license":null},{"id":"npm-codex","license":null},{"id":"npm-gemini-cli","license":null},{"id":"npm-claude-agent-sdk","license":null},{"id":"npm-openai-agents","license":null},{"id":"pypi-openai","license":null},{"id":"pypi-anthropic","license":null},{"id":"pypi-google-genai","license":null},{"id":"pypi-mistralai","license":null},{"id":"pypi-xai-sdk","license":null},{"id":"pypi-claude-agent-sdk","license":null},{"id":"pypi-openai-agents","license":null},{"id":"epoch-data-centers","license":null},{"id":"epoch-data-center-timelines","license":null}],"not_cleared":["openai-models-md","openai-pricing-md","anthropic-models-md","anthropic-pricing-md","xai-models-md","meta-models-md","meta-muse-wonder-md","meta-muse-spark-md","meta-pricing-md","deepseek-pricing-md","alibaba-pricing-html","alibaba-models-html","moonshot-models-md","openrouter-models","openai-status","anthropic-status","mistral-status","deepseek-status","deepseek-status-atom","xai-status","openai-misalignment-reports","cloudflare-changelog-ai","gpu-oracle","ramp-ai-index","github-prs-claude-code","github-prs-codex","github-prs-jules","limits-openai-gpt-6-astra","limits-openai-gpt-5-6-terra","limits-openai-gpt-5-6-luna","limits-anthropic","limits-xai","limits-meta","newsroom-openai","newsroom-meta"]},"columns":[{"name":"thesis_id","type":"text","meaning":"The thesis the claim belongs to. Its page is /theses/<thesis_id>.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"claim_id","type":"text","meaning":"The claim, unique across theses.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"date","type":"date","meaning":"The data date the reading is for.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"value","type":"number","meaning":"The measured value the claim’s kill condition is checked against. The thesis page gives the claim’s unit.","unit":"the claim’s own: a 0..1 share, a ratio, a growth rate, a count, USD, or USD per million tokens","nullable":false,"role":null,"as_of_only":false},{"name":"detail","type":"json","meaning":"JSON text (a string in the JSON export too): the inputs behind the value, shaped by the claim’s signal.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"computed_at","type":"timestamp","meaning":"When the reading was last computed with a different value or detail, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The source of the reading’s newest input.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that input was fetched, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"claim_verdicts","description":"The judge’s verdict on each candidate event for a judged thesis claim, with its rationale and the facts we computed (a price cut’s size, an outage’s hours). Rows are never edited.","csv":"/export/claim_verdicts.csv","json":"/export/claim_verdicts.json","key":["id"],"grain":"One row per candidate event the judge rated for a claim.","cadence":"When the judge runs.","time":{"event":"detected_at","knowledge":"judged_at"},"as_of":{"mode":"cut","note":"Verdicts with detected_at and judged_at both at or before as_of: a verdict is held back until it was judged."},"history":{"mode":"windowed","dated":["detected_at"],"full":"/export/claim_verdicts.csv?history=full","note":"Downloads the last 90 days by default, dated by detected_at, when we saw the change it rates. Add ?history=full for every row.","start":"From our first check of each source."},"joins":[{"column":"provider","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 42 are ours to license, 20 pass on under upstream terms (CC BY 4.0), and 35 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["google-pricing-html","google-models-md","mistral-models-md","mistral-pricing-md","openai-ashby","anthropic-greenhouse","anthropic-greenhouse-departments","xai-greenhouse","xai-greenhouse-departments","mistral-ashby","google-careers","edgar-fts","edgar-fts-openai","edgar-submissions-spcx","edgar-submissions-msft","edgar-submissions-amzn","edgar-submissions-nvda","edgar-submissions-googl","edgar-submissions-meta","edgar-submissions-orcl","edgar-submissions-baba","edgar-companyfacts-spcx","google-cloud-status","sec-revenue-net","sec-revenue-akam","sec-revenue-fsly","greenhouse-cloudflare","gpu-lambda","gpu-coreweave","gpu-nebius","gpu-crusoe","gpu-together","gpu-digitalocean","gpu-verda","gpu-azure","sec-revenue-crwv","sec-msft-openai","sec-amzn-anthropic","pay-openai","pay-anthropic","newsroom-google-deepmind","newsroom-mistral"],"upstream":[{"id":"openrouter-rankings-daily","license":"CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)"},{"id":"npm-wrangler","license":null},{"id":"npm-openai","license":null},{"id":"npm-anthropic-sdk","license":null},{"id":"npm-google-genai","license":null},{"id":"npm-mistral","license":null},{"id":"npm-claude-code","license":null},{"id":"npm-codex","license":null},{"id":"npm-gemini-cli","license":null},{"id":"npm-claude-agent-sdk","license":null},{"id":"npm-openai-agents","license":null},{"id":"pypi-openai","license":null},{"id":"pypi-anthropic","license":null},{"id":"pypi-google-genai","license":null},{"id":"pypi-mistralai","license":null},{"id":"pypi-xai-sdk","license":null},{"id":"pypi-claude-agent-sdk","license":null},{"id":"pypi-openai-agents","license":null},{"id":"epoch-data-centers","license":null},{"id":"epoch-data-center-timelines","license":null}],"not_cleared":["openai-models-md","openai-pricing-md","anthropic-models-md","anthropic-pricing-md","xai-models-md","meta-models-md","meta-muse-wonder-md","meta-muse-spark-md","meta-pricing-md","deepseek-pricing-md","alibaba-pricing-html","alibaba-models-html","moonshot-models-md","openrouter-models","openai-status","anthropic-status","mistral-status","deepseek-status","deepseek-status-atom","xai-status","openai-misalignment-reports","cloudflare-changelog-ai","gpu-oracle","ramp-ai-index","github-prs-claude-code","github-prs-codex","github-prs-jules","limits-openai-gpt-6-astra","limits-openai-gpt-5-6-terra","limits-openai-gpt-5-6-luna","limits-anthropic","limits-xai","limits-meta","newsroom-openai","newsroom-meta"]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of claim_id and subject: one verdict per claim and subject.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"thesis_id","type":"text","meaning":"The thesis the claim belongs to.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"claim_id","type":"text","meaning":"The judged claim.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"subject","type":"text","meaning":"What the verdict rates: the changes.id of a price cut, the entity_key of an incident or disclosure (so an incident updated twice is rated once), or the recipe_observations.id of a recipe item.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"change_id","type":"text","meaning":"The changes.id, or recipe_observations.id, of the record the judge was shown.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"provider","type":"text","meaning":"The company the event concerns: a lab id, or a company id such as cloudflare.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"verdict","type":"text","meaning":"The judge’s verdict, from the claim’s own set: competitive or generational for a price cut, material or not-material for an incident or disclosure. A recipe claim lists its own.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"rationale","type":"text","meaning":"The judge’s reason, at most 600 characters, grounded against the record and the facts.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"facts","type":"json","meaning":"JSON text (a string in the JSON export too): what the Worker computed about the event, such as a cut’s size or an outage’s hours.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"detected_at","type":"text","meaning":"When the rated record was detected: the change’s detected_at (an ISO-8601 timestamp), or for a recipe item the observation’s date (YYYY-MM-DD).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"judged_at","type":"timestamp","meaning":"When the judge run that wrote the verdict ran, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL the rated record was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that record was fetched, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"track_record_candidates","description":"Press articles (from GDELT) and lab newsroom posts found within 7 days either side of when we saw a notable or critical alert’s change. A candidate isn’t a match. track_record_verdicts says which ones report the same event. Rows are never edited.","csv":"/export/track_record_candidates.csv","json":"/export/track_record_candidates.json","key":["id"],"grain":"One row per article or newsroom post found for an alert.","cadence":"One search every 5 minutes, for the oldest alert whose window has closed.","time":{"event":"published_at","knowledge":"found_at"},"as_of":{"mode":"cut","note":"Rows with found_at at or before as_of."},"history":{"mode":"windowed","dated":["found_at"],"full":"/export/track_record_candidates.csv?history=full","note":"Downloads the last 90 days by default, dated by found_at, when we found the article. Add ?history=full for every row.","start":"From the first notable or critical alert."},"joins":[],"licensing":{"commercial":"upstream","note":"Under each source’s own license.","ours":[],"upstream":[{"id":"gdelt-doc","license":null}],"not_cleared":[]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of alert_id, kind and url: one row per alert, kind and URL.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"alert_id","type":"text","meaning":"The alerts.id the candidate was found for.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"kind","type":"text","meaning":"press, an article GDELT indexed, or newsroom, a post on the lab’s own newsroom feed.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"title","type":"text","meaning":"The headline as GDELT or the feed gives it. GDELT spaces out punctuation (“GPT - 6 . 1”), and we keep that as read.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"snippet","type":"text","meaning":"The first 400 characters of a newsroom post’s description. Null for press, since GDELT returns no article text.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"outlet","type":"text","meaning":"The site’s domain, as GDELT names it or from the post’s link.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"url","type":"text","meaning":"The article or post itself.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"published_at","type":"timestamp","meaning":"For press, when GDELT’s crawler first saw the article (not when the outlet published it). For a newsroom post, its own publication time. UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"search_id","type":"text","meaning":"The search that found it. Searches aren’t exported. The page shows each one’s outcome.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"found_at","type":"timestamp","meaning":"When the search that found it ran, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The GDELT query URL, or the newsroom feed the post was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that query or feed was fetched, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"track_record_verdicts","description":"The judge’s rating of each track-record candidate: whether it reports the alert’s own event, with a rationale grounded in the headline and the alert’s change rows. Rows are never edited.","csv":"/export/track_record_verdicts.csv","json":"/export/track_record_verdicts.json","key":["id"],"grain":"One row per candidate the judge rated.","cadence":"When the judge runs.","time":{"event":null,"knowledge":"judged_at"},"as_of":{"mode":"cut","note":"Rows with judged_at at or before as_of."},"history":{"mode":"windowed","dated":["judged_at"],"full":"/export/track_record_verdicts.csv?history=full","note":"Downloads the last 90 days by default, dated by judged_at, when the judge rated it. Add ?history=full for every row.","start":"From the first rated candidate."},"joins":[],"licensing":{"commercial":"upstream","note":"Under each source’s own license.","ours":[],"upstream":[{"id":"gdelt-doc","license":null}],"not_cleared":[]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of candidate_id: one verdict per candidate.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"candidate_id","type":"text","meaning":"The track_record_candidates.id rated.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"alert_id","type":"text","meaning":"The alert the candidate was found for.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"same_event","type":"boolean","meaning":"True when the judge read the headline (and snippet) as reporting the alert’s own event. A vague headline is rated false.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"rationale","type":"text","meaning":"The judge’s reason, at most 400 characters, grounded against the headline and the alert’s change rows.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"judged_at","type":"timestamp","meaning":"When the judge run that wrote the verdict ran, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The candidate’s own URL (the article or post rated).","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the candidate was fetched, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"recipe_observations","description":"What each thesis recipe read from its public source (layers/terminal/server/theses/recipes/). One row per recipe, key, and date, with the value, the fields kept beside it, and the R2 key of the raw bytes.","csv":"/export/recipe_observations.csv","json":"/export/recipe_observations.json","key":["id"],"grain":"One row per recipe, key, and date.","cadence":"Every 6 hours, with the full check of every source.","time":{"event":"date","knowledge":"first_seen_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. A recipe observation is rewritten in place when a re-read gives a different value, and the value it replaced is not kept."},"history":{"mode":"windowed","dated":["date"],"full":"/export/recipe_observations.csv?history=full","note":"Downloads the last 90 days by default, dated by date, the date the value is for. Add ?history=full for every row.","start":"Depends on the recipe. Some read a full history at their first run, others start at it."},"joins":[{"column":"company","to":"entity_identifiers.entity_id"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 19 are ours to license, 19 pass on under upstream terms, and 14 aren’t in any paid offer. A paid license covers only what comes from sources that are ours.","ours":["sec-revenue-net","sec-revenue-akam","sec-revenue-fsly","greenhouse-cloudflare","gpu-lambda","gpu-coreweave","gpu-nebius","gpu-crusoe","gpu-together","gpu-digitalocean","gpu-verda","gpu-azure","sec-revenue-crwv","sec-msft-openai","sec-amzn-anthropic","pay-openai","pay-anthropic","newsroom-google-deepmind","newsroom-mistral"],"upstream":[{"id":"npm-wrangler","license":null},{"id":"npm-openai","license":null},{"id":"npm-anthropic-sdk","license":null},{"id":"npm-google-genai","license":null},{"id":"npm-mistral","license":null},{"id":"npm-claude-code","license":null},{"id":"npm-codex","license":null},{"id":"npm-gemini-cli","license":null},{"id":"npm-claude-agent-sdk","license":null},{"id":"npm-openai-agents","license":null},{"id":"pypi-openai","license":null},{"id":"pypi-anthropic","license":null},{"id":"pypi-google-genai","license":null},{"id":"pypi-mistralai","license":null},{"id":"pypi-xai-sdk","license":null},{"id":"pypi-claude-agent-sdk","license":null},{"id":"pypi-openai-agents","license":null},{"id":"epoch-data-centers","license":null},{"id":"epoch-data-center-timelines","license":null}],"not_cleared":["cloudflare-changelog-ai","gpu-oracle","ramp-ai-index","github-prs-claude-code","github-prs-codex","github-prs-jules","limits-openai-gpt-6-astra","limits-openai-gpt-5-6-terra","limits-openai-gpt-5-6-luna","limits-anthropic","limits-xai","limits-meta","newsroom-openai","newsroom-meta"]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of recipe_id, key and date: one row per recipe, key and date.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"recipe_id","type":"text","meaning":"The recipe that read the value (layers/terminal/server/theses/recipes/).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"company","type":"text","meaning":"The company the recipe reads about: a lab id or a company id (cloudflare, coreweave, …).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"key","type":"text","meaning":"What the number is of: a revenue tag, a department, a package, a GPU offer, an entry.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"date","type":"date","meaning":"What the number is for: a quarter end, a day, an entry’s date, or the run’s UTC date for a source that reports a current state.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"value","type":"number","meaning":"The number read. Null where the entry carries none.","unit":"the recipe’s own (USD, USD per GPU-hour, a count, megawatts, …)","nullable":true,"role":null,"as_of_only":false},{"name":"payload","type":"json","meaning":"JSON text (a string in the JSON export too): the fields kept beside the value, and all a judged rationale may cite.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"raw_key","type":"text","meaning":"Key of the raw response the value was read from in this site’s object store (R2).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When this recipe, key and date was first read, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL the recipe fetched.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When the current value was read, in UTC. A re-read that finds the same value does not move it.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"filed_facts","description":"Every XBRL fact the exposure series read from Microsoft, Alphabet, Amazon, Meta, and Oracle: capex, finance leases, backlog, and what Microsoft and Amazon tag on OpenAI and Anthropic. One row per fact per filing, so a restated value keeps the row it replaced. Rows are never edited.","csv":"/export/filed_facts.csv","json":"/export/filed_facts.json","key":["id"],"grain":"One row per XBRL fact per filing.","cadence":"Every 30 minutes. Company facts every 6 hours.","time":{"event":"end","knowledge":"filed"},"as_of":{"mode":"cut","note":"Rows whose filing date is before the UTC day of as_of. A fact counts from the day after its filing, whenever we read it, so this table answers as_of back to 2019."},"history":{"mode":"windowed","dated":["end"],"full":"/export/filed_facts.csv?history=full","note":"Downloads the last 90 days by default, dated by end, the period end. Add ?history=full for every row.","start":"Company facts as far back as the filer tagged them, and each 10-Q and 10-K instance from 2020."},"joins":[{"column":"ticker","to":"entity_identifiers.ticker"},{"column":"cik","to":"entity_identifiers.cik"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["filed-facts-msft","filed-facts-googl","filed-facts-amzn","filed-facts-meta","filed-facts-orcl"],"upstream":[],"not_cleared":[]},"columns":[{"name":"id","type":"text","meaning":"SHA-256 (hex) of ticker, concept, dims, start, end, and accession_no.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"ticker","type":"text","meaning":"The filer: MSFT, GOOGL, AMZN, META, or ORCL.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"cik","type":"text","meaning":"The filer’s SEC CIK, ten digits, zero-padded.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"concept","type":"text","meaning":"The XBRL concept with its prefix, e.g. us-gaap:PaymentsToAcquirePropertyPlantAndEquipment.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"dims","type":"text","meaning":"The context’s dimensions as axis=member pairs, sorted and joined with “;”. A typed member’s value stands in for the member. Empty for a fact with no dimension.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"start","type":"date","meaning":"The period’s first day. Null for a balance, which is at end.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"end","type":"date","meaning":"The period’s last day, or the balance date.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"value","type":"number","meaning":"The value as tagged, signed as the filer tagged it.","unit":"the row’s unit column","nullable":false,"role":null,"as_of_only":false},{"name":"unit","type":"text","meaning":"USD, shares, pure (a ratio, 0.24 is 24%), or the measure as written.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"The filing’s form: 10-Q, 10-K, or an amendment.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"The filing date EDGAR gives the accession.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The filing’s EDGAR accession number.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"basis","type":"text","meaning":"companyfacts (EDGAR’s company facts, every fact with no dimension) or instance (the filing’s own XBRL instance, for a fact with dimensions).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"raw_key","type":"text","meaning":"Key of the document the fact was read from in this site’s object store (R2).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"first_seen_at","type":"timestamp","meaning":"When we first read the fact, in UTC.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The company facts document or the instance file the fact was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that read ran, in UTC. A filed fact is never rewritten.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"exposure_quarters","description":"The series on /exposure, one row per series and period: quarterly capex, backlog at each period end, and the OpenAI and Anthropic lines, from 2020. Each value is the newest filing’s. A derived quarter is a year-to-date figure less the one a quarter shorter, which the less_ columns name.","csv":"/export/exposure_quarters.csv","json":"/export/exposure_quarters.json","key":["series_id","end"],"grain":"One row per series and period.","cadence":"Rebuilt from filed_facts on each request.","time":{"event":"end","knowledge":"filed"},"as_of":{"mode":"cut","note":"Built from the filed_facts rows filed before the UTC day of as_of, so it answers back to 2019."},"history":{"mode":"windowed","dated":["end"],"full":"/export/exposure_quarters.csv?history=full","note":"Downloads the last 90 days by default, dated by end, the period end. Add ?history=full for every row.","start":"From 2020."},"joins":[{"column":"ticker","to":"entity_identifiers.ticker"}],"licensing":{"commercial":"yes","note":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":["filed-facts-msft","filed-facts-googl","filed-facts-amzn","filed-facts-meta","filed-facts-orcl"],"upstream":[],"not_cleared":[]},"columns":[{"name":"series_id","type":"text","meaning":"The series, e.g. msft-capex or amzn-anthropic-equity-securities-without-readily-determinable-fair-value-upward-price-adjustment-annual-amount.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"ticker","type":"text","meaning":"The filer: MSFT, GOOGL, AMZN, META, or ORCL.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filer","type":"text","meaning":"The filer’s name. Every value is the filer’s number, never a lab’s.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"group","type":"text","meaning":"capex, rpo (backlog), openai, or anthropic.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"label","type":"text","meaning":"What the series is, in plain words.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"concept","type":"text","meaning":"The XBRL concept the series reads.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"dims","type":"text","meaning":"The dimensions of the facts the series reads, as in filed_facts. Several are joined with “ | ” when the filer tagged the line on different axes over time.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"kind","type":"text","meaning":"flow (a value per quarter) or balance (a value at each period end).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"unit","type":"text","meaning":"USD, shares, or pure (a ratio).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"start","type":"date","meaning":"The quarter’s first day. Null for a balance.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"end","type":"date","meaning":"The quarter’s last day, or the balance date.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"value","type":"number","meaning":"The value for the quarter or at the date.","unit":"the row’s unit column","nullable":false,"role":null,"as_of_only":false},{"name":"derived","type":"boolean","meaning":"True when the quarter is a year-to-date figure less the one a quarter shorter, because the filing tags only year to date.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"less_start","type":"date","meaning":"For a derived quarter, the start of the figure taken away. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"less_end","type":"date","meaning":"For a derived quarter, the end of the figure taken away. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"less_value","type":"number","meaning":"For a derived quarter, the figure taken away. Null otherwise.","unit":"the row’s unit column","nullable":true,"role":null,"as_of_only":false},{"name":"less_accession_no","type":"text","meaning":"For a derived quarter, the filing the figure taken away came from. Null otherwise.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"accession_no","type":"text","meaning":"The filing the value is from: the newest that reported the period.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"form","type":"text","meaning":"That filing’s form.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filed","type":"date","meaning":"That filing’s date.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"filing_url","type":"text","meaning":"EDGAR’s index page for that filing.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The company facts document or the instance file the value was read from.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When that read ran, in UTC.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"source_runs","description":"One row per check of each source, with what we tried and what happened. There’s no fetched_at column, since a failed run fetched nothing.","csv":"/export/source_runs.csv","json":"/export/source_runs.json","key":["id"],"grain":"One row per check of a source.","cadence":"Model catalogs up to every 5 minutes, SEC feeds every 30, everything else every 6 hours.","time":{"event":null,"knowledge":"started_at"},"as_of":{"mode":"cut","note":"Rows with started_at at or before as_of."},"history":{"mode":"windowed","dated":["started_at"],"full":"/export/source_runs.csv?history=full","note":"Downloads the last 90 days by default, dated by started_at, when the check ran. Add ?history=full for every row.","start":"From our first check of each source."},"joins":[],"licensing":{"commercial":"yes","note":"Yes. These are records of our own checks. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone.","ours":[],"upstream":[],"not_cleared":[]},"columns":[{"name":"id","type":"text","meaning":"The run: <scope>:<source_id>:<started_at>.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"scope","type":"text","meaning":"Which poll ran it: edgar, survey, careers (the large careers feeds, on a six-hourly tick of their own), or tripwire (an off-cycle poll of the model catalogs).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"source_id","type":"text","meaning":"The source the row was read from, as sources.yaml names it. The Coverage panel lists each with its URL.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"started_at","type":"timestamp","meaning":"When the check started, in UTC. It stands in for fetched_at, since a failed run fetched nothing.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"status","type":"text","meaning":"ok, unchanged, failed, baseline (a first fetch, nothing to compare), skipped (the source needs a credential this deploy lacks), or absent (a URL derived from a stored row returned 404).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"detail","type":"text","meaning":"What happened, in words: the error, or a count such as “104 entities”.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"snapshot_id","type":"text","meaning":"The snapshots.id the run fetched. Null when no fetch succeeded.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"added","type":"integer","meaning":"Rows the run logged as added.","unit":"rows","nullable":false,"role":null,"as_of_only":false},{"name":"removed","type":"integer","meaning":"Rows the run logged as removed.","unit":"rows","nullable":false,"role":null,"as_of_only":false},{"name":"modified","type":"integer","meaning":"Rows the run logged as modified.","unit":"rows","nullable":false,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The URL attempted.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"entity_securities","description":"Each lab and SEC filer we track, mapped to the securities that carry it. Each row has the issuer, CIK, symbol, and exchange as EDGAR lists them, how the lab or filer relates to the security (itself, its parent, a disclosed investor), and the row’s source. A private lab with no security has a row saying so.","csv":"/export/entity_securities.csv","json":"/export/entity_securities.json","key":["entity","symbol"],"grain":"One row per tracked company and security that carries it (or none).","cadence":"By hand, when a mapping changes.","time":{"event":null,"knowledge":"checked_on"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. It’s a hand-checked table kept in code, and each row is dated by its checked_on."},"history":{"mode":"whole","dated":[],"full":null,"note":"Downloads whole, since it’s a hand-checked lookup of labs to securities.","start":"The current mapping only."},"joins":[{"column":"entity","to":"entity_identifiers.entity_id"},{"column":"cik","to":"entity_identifiers.cik"},{"column":"figi","to":"your security master"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 1 is ours to license and 1 passes on under upstream terms (FIGI identifiers are free for any use, commercial or not). A paid license covers only what comes from sources that are ours. figi and mic come from OpenFIGI. Everything else comes from SEC filings.","ours":["sec-identifiers"],"upstream":[{"id":"openfigi-mapping","license":"FIGI identifiers are free for any use, commercial or not (openfigi.com/docs/terms-of-service, read 2026-11-05)"}],"not_cleared":[]},"columns":[{"name":"entity","type":"text","meaning":"The lab or filer id (anthropic, microsoft, …), as the rest of the site uses it.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"entity_name","type":"text","meaning":"The lab’s or filer’s name.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_kind","type":"text","meaning":"lab (one of the nine) or filer (a company whose filings the site reads).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"relationship","type":"text","meaning":"self (the entity is the issuer or part of it), parent (a public parent consolidates it), investor (the issuer discloses a stake in it), or none (no security carries it).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"issuer","type":"text","meaning":"The issuer as EDGAR lists it. Null when no security carries the entity.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"cik","type":"text","meaning":"The issuer’s SEC CIK, ten digits. Null with no issuer.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"symbol","type":"text","meaning":"The symbol as EDGAR lists it, or the proposed one for a registered issuer. Null with no security.","unit":null,"nullable":true,"role":"key","as_of_only":false},{"name":"exchange","type":"text","meaning":"The exchange as EDGAR lists it. Null where no submissions feed is read.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"listing_status","type":"text","meaning":"listed (trading), registered (an S-1 on file, not yet trading), or private.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"note","type":"text","meaning":"How the mapping was made and its limits.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"mic","type":"text","meaning":"The ISO 10383 market identifier code OpenFIGI lists the symbol under: XNGS (Nasdaq Global Select Market) or XNYS (New York Stock Exchange). Null with no security.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"figi","type":"text","meaning":"The symbol’s US composite FIGI, from OpenFIGI. entity_identifiers has the URL and time it was read. Null with no security.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"disclosed_in","type":"text","meaning":"For a parent or investor row, the URL of the filing that discloses the relationship. Null for self and none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"The document the row was checked against.","unit":null,"nullable":false,"role":"provenance","as_of_only":false},{"name":"checked_on","type":"date","meaning":"The day the row was checked against source_url. It stands in for fetched_at, since the row is checked by hand.","unit":null,"nullable":false,"role":"provenance","as_of_only":false}]},{"name":"entity_identifiers","description":"One row per company we name: each lab, each SEC filer we read, and each comparison company. Each row has the legal name, country, CIK, LEI, and for a listed company its ticker, MIC, and FIGI. Every identifier has the URL and time we read it. entity_id is the value of every provider and company column, so this is the table to join on.","csv":"/export/entity_identifiers.csv","json":"/export/entity_identifiers.json","key":["entity_id"],"grain":"One row per company we name.","cadence":"By hand, when a company’s identifiers change.","time":{"event":null,"knowledge":"fetched_at"},"as_of":{"mode":"refused","note":"as_of answers 400 for this table. It’s a registry kept in code, and each identifier is dated by its own fetched_at."},"history":{"mode":"whole","dated":[],"full":null,"note":"Downloads whole, since it’s a lookup of one row per company.","start":"The current registry only."},"joins":[{"column":"cik","to":"your security master"},{"column":"lei","to":"your security master"},{"column":"figi","to":"your security master"},{"column":"ticker","to":"your security master"}],"licensing":{"commercial":"mixed","note":"It depends on the source. Of the sources it reads, 1 is ours to license, 2 pass on under upstream terms (CC0 1.0; FIGI identifiers are free for any use, commercial or not), and 1 isn’t in any paid offer. A paid license covers only what comes from sources that are ours. CIK, ticker, and exchange come from SEC, lei from GLEIF, and figi and mic from OpenFIGI. legal_name and country come from GLEIF too, except on a row with no LEI, where they come from the company’s own terms page.","ours":["sec-identifiers"],"upstream":[{"id":"gleif-lei","license":"CC0 1.0 (gleif.org/en/meta/lei-data-terms-of-use, read 2026-11-05)"},{"id":"openfigi-mapping","license":"FIGI identifiers are free for any use, commercial or not (openfigi.com/docs/terms-of-service, read 2026-11-05)"}],"not_cleared":["lab-legal-pages"]},"columns":[{"name":"entity_id","type":"text","meaning":"Our id for the company. It’s the value of provider, company, and entity_securities.entity everywhere else.","unit":null,"nullable":false,"role":"key","as_of_only":false},{"name":"entity_name","type":"text","meaning":"The name the site uses.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"entity_kind","type":"text","meaning":"lab (one of the nine), filer (a company whose SEC filings we read), or company (a comparison company in a thesis).","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"legal_name","type":"text","meaning":"The legal name as GLEIF records it, or as the company’s own terms page prints it when GLEIF has no record. Null where we read neither.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"country","type":"text","meaning":"ISO 3166-1 alpha-2 country of the legal entity: GLEIF’s legal-address country, else the country the company’s own page names. Null where we read neither.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"listing_status","type":"text","meaning":"listed (its own shares trade), registered (an S-1 on file, not yet trading), subsidiary (owned by a listed parent, which entity_securities names), or private.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"cik","type":"text","meaning":"The SEC CIK, ten digits, zero-padded. Null for a company that doesn’t file.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"cik_source_url","type":"text","meaning":"Where the CIK was read: SEC’s company_tickers_exchange.json.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"cik_fetched_at","type":"timestamp","meaning":"When we read the CIK, in UTC.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"lei","type":"text","meaning":"The ISO 17442 Legal Entity Identifier, from GLEIF. Null with the reason in lei_note.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"lei_status","type":"text","meaning":"GLEIF’s registration status for the LEI (ISSUED, LAPSED, …), as of lei_fetched_at.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"lei_note","type":"text","meaning":"How the LEI was matched to the company, or why there’s none.","unit":null,"nullable":false,"role":null,"as_of_only":false},{"name":"lei_source_url","type":"text","meaning":"The GLEIF request the LEI was read from. For a null LEI, the search that found none, where we ran one.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"lei_fetched_at","type":"timestamp","meaning":"When that GLEIF request ran, in UTC.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"ticker","type":"text","meaning":"The primary symbol as EDGAR lists it, or the proposed one for a registered company. entity_securities lists every symbol we map. Null when not listed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"exchange","type":"text","meaning":"The exchange as EDGAR’s ticker file lists it. Null when EDGAR gives none.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"mic","type":"text","meaning":"The ISO 10383 market identifier code OpenFIGI lists the symbol under (XNGS or XNYS). Null when not listed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"figi","type":"text","meaning":"The symbol’s US composite FIGI, from OpenFIGI. Null when not listed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"share_class_figi","type":"text","meaning":"The share-class FIGI OpenFIGI gives with it. Null when not listed.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"figi_source_url","type":"text","meaning":"The OpenFIGI mapping API the FIGIs and MIC were read from.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"figi_fetched_at","type":"timestamp","meaning":"When that OpenFIGI request ran, in UTC.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"note","type":"text","meaning":"Anything a join needs to know.","unit":null,"nullable":true,"role":null,"as_of_only":false},{"name":"source_url","type":"text","meaning":"Where legal_name and country were read: GLEIF when there’s an LEI, else the company’s own terms page.","unit":null,"nullable":true,"role":"provenance","as_of_only":false},{"name":"fetched_at","type":"timestamp","meaning":"When source_url was read, in UTC.","unit":null,"nullable":true,"role":"provenance","as_of_only":false}]}],"sources":[{"id":"openai-models-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"openai-pricing-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"anthropic-models-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"anthropic-pricing-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"xai-models-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"google-pricing-html","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"google-models-md","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"meta-models-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"meta-muse-wonder-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"meta-muse-spark-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"meta-pricing-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"mistral-models-md","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"mistral-pricing-md","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"deepseek-pricing-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"alibaba-pricing-html","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"alibaba-models-html","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"moonshot-models-md","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"openrouter-models","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"openrouter-rankings-daily","group":"sources","rights":"upstream","license":"CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)","commercial":"Under its own license: CC BY 4.0 (openrouter.ai/docs/cookbook/administration/data-api, read 2026-09-08)."},{"id":"openai-ashby","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"anthropic-greenhouse","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"anthropic-greenhouse-departments","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"xai-greenhouse","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"xai-greenhouse-departments","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"mistral-ashby","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"google-careers","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-fts","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-fts-openai","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-spcx","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-msft","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-amzn","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-nvda","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-googl","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-meta","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-orcl","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-submissions-baba","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"edgar-companyfacts-spcx","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"openai-status","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"anthropic-status","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"google-cloud-status","group":"sources","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"mistral-status","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"deepseek-status","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"deepseek-status-atom","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"xai-status","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"openai-misalignment-reports","group":"sources","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"sec-revenue-net","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"sec-revenue-akam","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"sec-revenue-fsly","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"npm-wrangler","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"greenhouse-cloudflare","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"cloudflare-changelog-ai","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"gpu-lambda","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-coreweave","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-nebius","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-crusoe","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-together","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-digitalocean","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-verda","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-azure","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gpu-oracle","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"sec-revenue-crwv","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"ramp-ai-index","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"npm-openai","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-anthropic-sdk","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-google-genai","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-mistral","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-claude-code","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-codex","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-gemini-cli","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-claude-agent-sdk","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"npm-openai-agents","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-openai","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-anthropic","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-google-genai","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-mistralai","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-xai-sdk","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-claude-agent-sdk","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"pypi-openai-agents","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"github-prs-claude-code","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"github-prs-codex","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"github-prs-jules","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"epoch-data-centers","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"epoch-data-center-timelines","group":"recipes","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"sec-msft-openai","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"sec-amzn-anthropic","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"limits-openai-gpt-6-astra","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"limits-openai-gpt-5-6-terra","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"limits-openai-gpt-5-6-luna","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"limits-anthropic","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"limits-xai","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"limits-meta","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"pay-openai","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"pay-anthropic","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"newsroom-openai","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"newsroom-google-deepmind","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"newsroom-meta","group":"recipes","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"newsroom-mistral","group":"recipes","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gdelt-doc","group":"track_record","rights":"upstream","license":null,"commercial":"Under its own license, which we pass on."},{"id":"wayback-pricing","group":"archive","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"wayback-hiring","group":"archive","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."},{"id":"filed-facts-msft","group":"filed_facts","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"filed-facts-googl","group":"filed_facts","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"filed-facts-amzn","group":"filed_facts","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"filed-facts-meta","group":"filed_facts","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"filed-facts-orcl","group":"filed_facts","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"sec-identifiers","group":"identifiers","rights":"ours","license":null,"commercial":"Yes. A paid license covers the full history. The last 90 days are CC BY 4.0 for anyone."},{"id":"gleif-lei","group":"identifiers","rights":"upstream","license":"CC0 1.0 (gleif.org/en/meta/lei-data-terms-of-use, read 2026-11-05)","commercial":"Under its own license: CC0 1.0 (gleif.org/en/meta/lei-data-terms-of-use, read 2026-11-05)."},{"id":"openfigi-mapping","group":"identifiers","rights":"upstream","license":"FIGI identifiers are free for any use, commercial or not (openfigi.com/docs/terms-of-service, read 2026-11-05)","commercial":"Under its own license: FIGI identifiers are free for any use, commercial or not (openfigi.com/docs/terms-of-service, read 2026-11-05)."},{"id":"lab-legal-pages","group":"identifiers","rights":"not-cleared","license":null,"commercial":"No. It stays on the free site and isn’t in any paid offer."}],"changelog":[{"version":"1.5.1","date":"2026-11-05","summary":"The Parquet bulk files are compressed with gzip instead of zstd, which failed on the largest tables. Readers pick the codec up from each file. No columns changed.","tables":[]},{"version":"1.5.0","date":"2026-11-05","summary":"Hiring history from before we read the labs’ job boards, from the Internet Archive’s copies of each lab’s own board: a new archived_hiring table (open roles per copy, on the whole board and per department), and an axis column on archive_captures saying which table a copy’s rows went to.","tables":[{"name":"archived_hiring","added":["id","capture_id","provider","scope","label","department_id","department","open_roles","captured_at","original_url","basis","source_url","fetched_at"],"removed":[],"renamed":[],"note":null},{"name":"archive_captures","added":["axis"],"removed":[],"renamed":[],"note":"Every existing row is pricing."}]},{"version":"1.4.0","date":"2026-11-05","summary":"Security-master identifiers. entity_identifiers has one row per company we name, with its CIK, LEI, ticker, MIC and FIGI, each with the URL and time we read it. Join any provider or company column to its entity_id. entity_securities gains mic, figi, and disclosed_in, the filing that discloses a parent or an investor. It also gains a row for Alphabet’s stake in Anthropic, from Anthropic’s S-1/A.","tables":[{"name":"entity_identifiers","added":["entity_id","entity_name","entity_kind","legal_name","country","listing_status","cik","cik_source_url","cik_fetched_at","lei","lei_status","lei_note","lei_source_url","lei_fetched_at","ticker","exchange","mic","figi","share_class_figi","figi_source_url","figi_fetched_at","note","source_url","fetched_at"],"removed":[],"renamed":[],"note":null},{"name":"entity_securities","added":["mic","figi","disclosed_in"],"removed":[],"renamed":[],"note":"A new investor row: anthropic to GOOGL, disclosed in Anthropic’s S-1/A."}]},{"version":"1.3.0","date":"2026-11-05","summary":"New: every bulk file also as Parquet (zstd), the same rows and columns typed by the data dictionary, listed under parquet in each manifest entry. Licensees can also read the full-history files over S3 with read-only credentials. No table columns changed.","tables":[]},{"version":"1.2.0","date":"2026-11-04","summary":"Two tables for the price history from before we watched the labs, read from the Internet Archive’s copies of each lab’s own pricing page: archived_prices (one row per SKU per copy) and archive_captures (every copy considered, read or skipped with the reason).","tables":[{"name":"archived_prices","added":["id","capture_id","entity_key","provider","content_hash","payload","captured_at","original_url","basis","source_url","fetched_at"],"removed":[],"renamed":[],"note":null},{"name":"archive_captures","added":["id","era","provider","original_url","captured_at","status","detail","content_hash","raw_key","bytes","rows","source_url","fetched_at"],"removed":[],"renamed":[],"note":null}]},{"version":"1.1.0","date":"2026-11-04","summary":"New: alert webhooks for licensees. Each POST carries schema_version, the alert with the columns of the alerts table plus permalink, source_urls and its cited rows of the changes table, and the license for your key. No table columns changed.","tables":[]},{"version":"1.0.0","date":"2026-11-04","summary":"First versioned schema. Every export, bulk file, and API response states its schema_version from here on. A full-history export adds a license column before source_url, and a rebuilt export (?as_of=) adds as_of_verified. Neither changes the default columns.","tables":[]}]}