OpenAI
no audited disclosureNo audited disclosure yet.No SEC filing on record carries this lab’s revenue. Press-reported run-rates are deliberately not shown: a figure this terminal cannot link to the document it came from is a claim, not data.
Frontier Terminal
Tracking the frontier AI labs — API prices, hiring, filings, revenue, incidents, demand share. Every number carries its source URL and fetch timestamp.
Signal
newest fetch
- notable
OpenAI's Go To Market department shows 5 job removals against 4 additions in this batch, including three "Account Director" postings removed.
In the records reviewed, OpenAI's Go To Market department had 5 roles removed — "Client Partner - Media & Entertainment, Ads Solutions (Los Angeles, CA)", "Account Director, Tech", "Account Director, Media", "Partnership Sales Manager", and "Account Director, GSIs & Consulting" — versus 4 roles added — "Account Director, Large Enterprise", "Partner Solutions Engineer, Ads Solutions - San Francisco", "Solutions Engineer, Ads Solutions - San Francisco", and "Demo Experience Engineer, Technical Success". Three of the five removed titles are "Account Director" roles on the Sales team, while two of the four added roles are Ads Solutions-team "Solutions Engineer" titles.
- notable
xAI adds nine "Data Center" roles in one batch, all Memphis power-generation positions
Nine new postings landed at once under xAI's "Data Center" department, all located in Southaven, MS; Memphis, TN: "Electrical Engineer (Power Generation) - Memphis", "Operations Analyst (Power Generation) - Memphis", "Turbine Technician - Memphis", "Manager, Engineering (Power Generation & Utilities) - Memphis", "Fluids Engineer (Power Generation) - Memphis", "Controls Engineer (Power Generation) - Memphis", "Plant Technician (Power Generation) - Memphis", "Process Technician (Fluids) - Memphis", and "Lead Engineer (Power Generation) - Memphis". Nine adds in one department clears the five-or-more hiring-shift threshold and reads as a power-generation buildout at the Memphis site.
- notable
xAI launches grok-4.7 with new "standard" and "long_context" pricing tiers
xAI added two new model rows for model_slug "grok-4.7": the "standard" tier lists input at 2 per MTok, cached input at 0.5 per MTok, output at 6 per MTok, with a 500k context window; the "long_context" tier lists input at 4 per MTok, cached input at 1 per MTok, output at 12 per MTok, also with a 500k context window. Both are new USD-priced SKUs, not changes to an existing row.
- notable
SPACE EXPLORATION TECHNOLOGIES CORP (CIK 0001181412) files SEC Form 4, accession 0001213900-26-100186, dated 2026-09-15.
EDGAR records a new Form 4 filing for SPACE EXPLORATION TECHNOLOGIES CORP (CIK 0001181412), accession 0001213900-26-100186, filed 2026-09-15.
- notable
Google adds two new Gemini Live model SKUs and re-tiers gemini-3.1-flash-live-preview
Google's catalog adds two new SKUs, gemini-3.8-live and gemini-3.8-live-extended-thinking (tier standard), both priced at $0.75 input / $4.50 output per MTok (text); the untiered gemini-3.1-flash-live-preview record is removed and replaced by a tier: standard version of the same slug at the same $0.75 input / $4.50 output per MTok pricing.
- notable
Anthropic status page: major-impact incident hit Claude API (api.anthropic.com), Claude Code, Claude Cowork, and claude.ai, now resolved
An incident titled "Elevated errors for Claude Mythos 5.1 and Claude Fable 5.1" listed components Claude API (api.anthropic.com), Claude Code, Claude Cowork, and claude.ai with impact major; status is resolved.
- notable
OpenAI adds new model SKU gpt-live-1
OpenAI's model list adds gpt-live-1, described as "Our premier model for natural, expressive voice conversations with smooth interruption handling." No input, cached input, or output price per MTok is listed yet for this SKU.
- notable
Google restructures Gemini Robotics-ER 2 pricing: lower through December 31, 2026, then reverting January 1, 2027.
Gemini Robotics-ER 2 (streaming-preview and preview, standard and batch tiers) prices per MTok fall through December 31, 2026 then rise back to prior levels starting January 1, 2027: streaming-preview standard input falls from $2.00 to $1.00 (back to $2.00), output from $10.00 to $5.00 (back to $10.00); preview standard input from $2.00 to $1.00 (back to $2.00), cached_input from $0.20 to $0.10 (back to $0.20), output from $10.00 to $5.00 (back to $10.00); preview batch input from $1.00 to $0.50 (back to $1.00), cached_input from $0.10 to $0.05 (back to $0.10), output from $5.00 to $2.50 (back to $5.00).
- notable
Google delists three model price lines: gemini-embedding-001, gemini-robotics-er-1.6-preview, and gemini-robotics-er-2-streaming-preview
Google removed all price rows for gemini-embedding-001 (standard and batch tiers), gemini-robotics-er-1.6-preview (standard and batch tiers), and gemini-robotics-er-2-streaming-preview (standard tier, including its effective 2027-01-01 row) from its published price list.
- notable
OpenAI adds two new image-generation model SKUs: gpt-image-2.5-flare and gpt-image-2.5-sunburst.
New model rows appear for gpt-image-2.5-flare, described as "Fast, high-quality everyday image generation", and gpt-image-2.5-sunburst, described as "Our most capable model for image generation and editing"; no pricing is published yet for either.
Ticker
- xAI: 5 Greenhouse job-board changes across Safety, Data Center, and G&A.boards-api.greenhouse.io
- Anthropic: 4 Greenhouse job-board changes across Technical Education, Safeguards (Trust & Safety), and Product Management, Support, & Operations.boards-api.greenhouse.io
- OpenAI: 17 further job-board changes outside the Go To Market shift, including a 3-role "Operating Systems Engineer" cluster added in Consumer Devices.api.ashbyhq.com
- xAI removes one "Data Center" role and updates another amid the Memphis buildoutboards-api.greenhouse.io
- Anthropic adds three roles across Finance and Sales, removes one Sales role, updates one Infrastructure postingboards-api.greenhouse.io
- OpenAI adds five roles across Go To Market, Applied AI and Finance, removes six across six departmentsapi.ashbyhq.com
- Anthropic adds four roles: three in Sales (London, UK; Bangalore, India; Singapore) and one in Applied AI (Mumbai, India).boards-api.greenhouse.io
- Anthropic hiring: adds "Software Engineer, Beneficial Deployments" and removes "Engineering Manager, Labs", both in Engineering & Design - Productboards-api.greenhouse.io
88 changes in the 24h to pricing 2hiring 84sec 2incidents 0
Like-for-like API pricing
$ per million tokens, on each vendor’s own flagship / balanced / economy recommendation.
| Provider | Flagshipthe vendor’s own “start here” model for hard work | Balancedthe vendor’s stated intelligence-for-cost tradeoff | Economythe vendor’s model for cost-sensitive, high-volume work |
|---|---|---|---|
| OpenAI | |||
| Anthropic | |||
not mapped | not mapped | not mapped | |
| xAI | not mapped | not mapped |
List price on each vendor’s standard tier — batch, long-context, promo-window and cached-only rows are excluded (they are in the full catalog). Class assignment is the one editorial judgment on this page: a model sits in a column because the vendor’s own docs recommend it there — open the info control beside a model for the sentence and its source. Prices, timestamps and URLs are read from the store untouched.
Disclosed revenue
What each lab, or the public parent that consolidates it, has reported to the SEC — and an explicit blank where nothing audited exists.
Anthropic
no audited disclosureNo audited disclosure yet.No SEC filing on record carries this lab’s revenue. Press-reported run-rates are deliberately not shown: a figure this terminal cannot link to the document it came from is a claim, not data.
No audited disclosure yet.No SEC filing on record carries this lab’s revenue. Press-reported run-rates are deliberately not shown: a figure this terminal cannot link to the document it came from is a claim, not data.
xAI
parent filingParent’s consolidated revenue
SPACE EXPLORATION TECHNOLOGIES CORP. (SPCX) consolidated — the parent’s top line, not xAI’s
Primary source only: audited SEC disclosures read from EDGAR’s XBRL company facts, each value linked to the filing that reported it. No audited disclosure on file means nothing is shown — press-reported run-rates are deliberately not shown.
Demand share
Each lab’s share of tokens routed through OpenRouter in the newest seven days — a proxy for relative demand while the labs are private.
Share of tokens routed, 2026-09-15 → 2026-09-21 · change vs 2026-09-08 → 2026-09-14
- OpenAI10.5%−7.1 pp
- Anthropic3.0%−0.3 pp
- Google4.6%−0.7 pp
- xAI0.2%−0.1 pp
- Other81.8%+8.1 pp
30-day series and top models per lab
OpenRouter share, not market share — one aggregator’s traffic
OpenRouter usage rankings (daily) · dataset as of
Also published, not ingested (license or no data endpoint — see coverage): Artificial Analysis, LMArena (arena.ai)
Implied spend share
The same tokens at each vendor’s list price: where the dollars would go if the channel paid list — a range, with the labs it could not price named.
Tokens routed 2026-09-15 → 2026-09-21, at each vendor’s list price as published — implied, not observed.
| Lab | Spend share | Implied $ | Token share | Priced |
|---|---|---|---|---|
| OpenAI | not priceable — see below | — | 57.5% · 13.8T | 0.0% |
| Anthropic | 82.1% | $16.8M–$84.0M | 16.2% · 3.89T | 100.0% |
14.9%–16.1% | $3.05M–$16.5M | 25.0% · 6.01T | 89.9% | |
| xAI | 1.8%–3.0% | $606K–$1.82M | 1.3% · 303B | 100.0% |
In the token share, out of the dollar share
OpenAI: listed SKUs carry no published price on the vendor page. The spend share is among the labs that could be priced, and says nothing about the rest.
108T of 132T tokens in the window went to models outside the four labs (open-weight and others) and are not priced here — the “tokens to open source, economics to the frontier labs” split this table exists to show.
The join per model, and what could not be priced
Implied OpenRouter-channel spend at list price — not revenue, not market share
OpenRouter usage rankings (daily)OpenAI models.mdAnthropic models overview.mdAnthropic pricing.mdxAI models.mdGoogle Gemini API pricing
Recent releases
SKUs that appeared on a vendor page already being watched, with the price printed beside them that day.
- $4.00 in$12.00 out
- $2.00 in$6.00 out
- $0.75 in$4.50 out
- $0.75 in$4.50 out
- $0.75 in$4.50 out
Hiring mix
Open roles by department, from each board’s current listing, and the change over the comparison window.
OpenAI
813 open roles · +59 in 28d- Go To Market148 18%
- Scaling118 15%
- Applied AI105 13%
- Research68 8%
- User Operations53 7%
- Marketing47 6%
- Security32 4%
- Consumer Devices26 3%
+ 24 more departments
Anthropic
617 open roles · +73 in 28d- Sales123 20%
- AI Research & Engineering67 11%
- Finance52 8%
- Applied AI50 8%
- Security50 8%
- Engineering & Design - Product46 7%
- Software Engineering - Infrastructure43 7%
- Safeguards (Trust & Safety)40 6%
+ 12 more departments
company_name on the rows: Anthropic
No public feed
None usable: careers.google.com is an auth-walled SPA; the DeepMind Greenhouse board is vestigial (10 jobs). Alphabet-wide numbers would conflate Search/Ads/Cloud with Gemini and mislead. Verdict transcribed from the source audit — showing nothing beats guessing.
xAI
275 open roles · +20 in 28dEntity-blended board
262 roles at audit time; 255 at the 2026-08-25T23:08Z join re-fetch. Every row carries company_name "SpaceXAI" — the board is an entity-blended SpaceX/xAI artifact of the 2025 merger, and presenting it as pure xAI hiring would mislead. Caveat must ship visibly wherever this data surfaces. What the /departments join adds is bounded evidence, not a fix: the 26 departments are xAI-shaped (Data Center 121, Human Data 36, Engineering 23) with no aerospace or launch function, and 2 of 255 titles name a SpaceX-side program (Starlink, battery storage). Department requires the join — the jobs endpoint carries no department field.
- Data Center140 51%
- Engineering26 9%
- Human Data23 8%
- Product14 5%
- Safety11 4%
- Information Security8 3%
- Model8 3%
- Sales8 3%
+ 11 more departments
company_name on the rows: SpaceXAI
Capacity buildout
Open roles in each board’s physical-infrastructure departments — data center, facilities, energy, construction, compute, hardware — the supply side of the strain panel below.
OpenAI
0 open · 0 since 2026-08-25No department on this board matches the cluster.The board’s own department labels name none of the cluster terms; a zero here is a label fact, not a hiring one.
Anthropic
26 open · -7 since 2026-08-254% of the board · summed from the board’s own labels:
- Compute24 -7
- Hardware2
No public feed
None usable: careers.google.com is an auth-walled SPA; the DeepMind Greenhouse board is vestigial (10 jobs). Alphabet-wide numbers would conflate Search/Ads/Cloud with Gemini and mislead. Verdict transcribed from the source audit — showing nothing beats guessing.
xAI
145 open · +145 since 2026-08-2553% of the board · summed from the board’s own labels:
- Data Center140 +140
- Infrastructure5 +5
Entity-blended board
262 roles at audit time; 255 at the 2026-08-25T23:08Z join re-fetch. Every row carries company_name "SpaceXAI" — the board is an entity-blended SpaceX/xAI artifact of the 2025 merger, and presenting it as pure xAI hiring would mislead. Caveat must ship visibly wherever this data surfaces. What the /departments join adds is bounded evidence, not a fix: the 26 departments are xAI-shaped (Data Center 121, Human Data 36, Engineering 23) with no aerospace or launch function, and 2 of 255 titles name a SpaceX-side program (Starlink, battery storage). Department requires the join — the jobs endpoint carries no department field.
A department belongs to the cluster when its name, as the board prints it, contains one of the cluster’s terms (Data Center, Infrastructure, Facilities, Energy, Construction, Compute, Hardware) and not “Software”. Terms live in the source registry, and the judge reads the same list: a five-role move across these departments at one lab is a capacity signal, not a hiring one.
Capacity strain
Incidents each lab posted on its own status page: the newest 30 days against the 30 before, and what is open now.
OpenAI
0 open now- last 30d32
- prior 30d15
Held history begins Aug 5 16:55Z — the prior window is partial until the feed has been polled through it.
Anthropic
0 open now- last 30d17
- prior 30d31
- last 30d0
- prior 30d0
xAI
No public feed
status.x.ai/api/v2/incidents.json answers 403 (Cloudflare challenge page) to any non-browser client, live-checked 2026-09-08 with a descriptive User-Agent. Scraping a rendered page would be a dependency on a bot wall; the terminal shows the gap instead. Verdict transcribed from the source audit — showing nothing beats guessing.
Windows count incidents by the start time the vendor posted, back from the newest status fetch (). Impact and status are each vendor’s own words — Statuspage’s none / minor / major / critical is not mapped onto Google’s low / medium / high.
Coverage
Every source behind the numbers above: newest snapshot, last poll, and the audit’s caveat.
| Source | Newest snapshot | Last poll | Snapshots | Rows | Caveat |
|---|---|---|---|---|---|
| OpenAI models.mdOpenAI · pricing-catalog | unchanged | 99 | 100 | ||
| OpenAI pricing.mdOpenAI · pricing-catalog | never fetched | no run | 0 | 0 | |
| Anthropic models overview.mdAnthropic · pricing-catalog | unchanged | 99 | 4 | ||
| Anthropic pricing.mdAnthropic · pricing-catalog | unchanged | 99 | 34 | ||
| xAI models.mdxAI · pricing-catalog | ok | 97 | 16 | ||
| Google Gemini API pricingGoogle · pricing-catalog | ok | 95 | 94 | ||
| OpenRouter /api/v1/modelsmarket-wide · pricing-cross-check · cross-check only — never a source of record | ok | 98 | 0 | ||
| OpenRouter usage rankings (daily)market-wide · demand-share | ok | 51 | 2193 | ||
| OpenAI Ashby job boardOpenAI · hiring | ok | 99 | 813 | ||
| Anthropic Greenhouse jobsAnthropic · hiring | ok | 98 | 617 | ||
| Anthropic Greenhouse departments (join)Anthropic · hiring · join table — fetched to enrich another source, yields no rows of its own | ok | 98 | 0 | — | |
| xAI Greenhouse jobsxAI · hiring | ok | 99 | 275 | ||
| xAI Greenhouse departments (join)xAI · hiring · join table — fetched to enrich another source, yields no rows of its own | ok | 94 | 0 | ||
| EDGAR full-text search (S-1 tripwire, “Anthropic”)market-wide · sec | ok | 1084 | 54 | ||
| EDGAR full-text search (S-1 tripwire, “OpenAI”)market-wide · sec | never fetched | no run | 0 | 0 | |
| EDGAR submissions (SPCX — xAI’s parent)market-wide · sec | unchanged | 1090 | 82 | ||
| EDGAR submissions (MSFT)market-wide · sec | unchanged | 2 | 81 | ||
| EDGAR submissions (AMZN)market-wide · sec | unchanged | 2 | 86 | ||
| EDGAR submissions (NVDA)market-wide · sec | ok | 2 | 87 | ||
| EDGAR submissions (GOOGL)market-wide · sec | unchanged | 2 | 54 | ||
| EDGAR XBRL company facts (SPCX — xAI’s parent)market-wide · revenue | unchanged | 2 | 4 | ||
| OpenAI status page incidentsOpenAI · status | unchanged | 52 | 47 | ||
| Anthropic (Claude) status page incidentsAnthropic · status | ok | 52 | 55 | ||
| Google Cloud status incidents (Gemini / Vertex AI only)Google · status | unchanged | 52 | 1 |
Not tracked, on purpose
- google-hiring — cut — display "no public feed" honestly. None usable: careers.google.com is an auth-walled SPA; the DeepMind Greenhouse board is vestigial (10 jobs). Alphabet-wide numbers would conflate Search/Ads/Cloud with Gemini and mislead.
- artificial-analysis — cut — external link only, never ingested. artificialanalysis.ai/data-api (read 2026-09-08) states the free API tier is "Internal use only; no redistribution." and that redistribution rights come only with a commercial package ("Commercial redistribution with attribution"). A public terminal republishes by definition, so the site is linked as a reference beside the demand-share panel and no figure from it is stored or shown.
- lmarena — cut — external link only, never ingested. arena.ai (formerly lmarena.ai; read 2026-09-08) publishes a leaderboard app with no documented data endpoint, dataset download, or data license. Scraping a page with no stated license would be a claim about that page, not data; it is linked as a reference beside the demand-share panel.
- compute-deals — stretch goal only (new-post tier, no extraction claims). RSS detection solvable for 4 sources (Anthropic = SPA no feed; xAI = hard 403), but the deals that matter break in the press before any company feed — honest coverage is impossible, so extraction is cut.
- xai-status — cut — display "no public feed" honestly. status.x.ai/api/v2/incidents.json answers 403 (Cloudflare challenge page) to any non-browser client, live-checked 2026-09-08 with a descriptive User-Agent. Scraping a rendered page would be a dependency on a bot wall; the terminal shows the gap instead.
- mistral — cut — not tracked on any axis. No supported surface on either primary axis, live-checked 2026-09-08. Hiring: api.ashbyhq.com/posting-api/job-board/mistral answers 404 (the board at jobs.ashbyhq.com/mistral is a 7 KB HTML shell over an internal GraphQL endpoint, not the posting API). Pricing: mistral.ai/pricing is a 485 KB client-side HTML app (same bytes at docs.mistral.ai/deployment/laplateforme/pricing) with no markdown or JSON endpoint; docs.mistral.ai/llms.txt lists a Pricing and a Models Overview .md page, but both links answer 404 with a 462 KB HTML shell. An HTML parse in the Google style is conceivable; with no hiring feed to pair it with, a fifth lab on one axis is not worth the parser.
Every label links to the exact URL fetched; the age beside it carries the exact UTC timestamp. The info control on a row opens the audit’s caveat for that source, in its own words.