Skip to Sidebar Skip to Content
Company tracker 1 of 10 · 10 companies, one lens

OpenAI

The maker of ChatGPT and the GPT models, tracked through the same six lenses as every company on AI Adoption.
Its best model, GPT-6 Astra, scores 168.8 on the Epoch Capabilities Index, the highest of the 10 tracked companies. Annualized revenue reached $50B by 30 Sep 2026, and its Mar 2026 round valued it at $852.0B. In March 2026, 31.0% of US adults said they had used ChatGPT in the past week. It released 21 notable models in 2025, the most of the tracked companies.
Follow OpenAICompare with the others
65 ledger entries33 charts name OpenAILatest entry Sep 2026Checked 20 Oct 2026

OpenAI through six lenses

How the lenses work →
The same six lenses, in the same order, on every company page. Each shows one comparable number where a source measures several companies the same way, its rank, and when it last changed.
01Capability#1 of 10
How good are its best models?
168.8
Epoch Capabilities Index score of its best model, GPT-6 Astra (Sep 2026)
Anthropic168.8OpenAI168.8Google166.4Meta161.8
Epoch combines many benchmarks into one scale. 90% interval for this model: 165.4 to 174.5.
Text Arena, Mar 2026: 1,481, #4 of 8. Arena chart →
Independent measureChanged Sep 2026Epoch AI data →
02Output#1 of 10
How much does it ship?
21
Notable AI models released in 2025
OpenAI21Google15Alibaba11Anthropic7
61 notable models from 2014 to 2025 and 18 in 2026 so far. Model record →
Independent measureChanged 2025View chart →
03Money#2 of 9
How much does it earn?
$50B
Annualized revenue, 30 Sep 2026
Anthropic$65BOpenAI$50BZ.ai (Zhipu)$1.6BDeepSeek$1B
Annualized run rate; Epoch rates the report “Likely”. Dates differ by company; each is its latest report. Ranked among the 9 companies in Epoch AI's revenue data.
Valuation $852.0B after its Mar 2026 round; $182.8B raised in equity in all; about 4,500 staff (Mar 2026).
Reported figureChanged Sep 2026Epoch AI data →
04Compute#1 of 10
How much compute and energy goes into its AI?
9.4×10²⁶
FLOP: its largest training run with a published estimate, GPT-6 Astra (Sep 2026)
OpenAI9.4×10²⁶xAI5×10²⁶Meta5.2×10²⁵Google5×10²⁵
Largest estimated training cost: $366M (GPT-4.5), Epoch AI. Model record →
Ranked among the 10 tracked companies.
Reported compute spend
OpenAI spend, 2025$16.3BAnthropic spend, 2025$6.8B
R&D Inference. Epoch AI reports spend for these two labs only.
Had about 1.74M H100-equivalents available at the end of 2025, #1 of 5 labs. Epoch AI data →
Independent estimateChanged Sep 2026View chart →
05Use#1 of 7
Who uses it, and for what?
31.0%
of US adults said they used ChatGPT in the past week, March 2026
OpenAI31.0%Google20.9%Microsoft10.5%Meta7.5%
Epoch AI with Ipsos, a probability-based sample of 2,021 US adults; 49.2% used no AI service. Among employed adults in July 2026: 33.7%.
14,376 US AI job postings named ChatGPT as a skill in 2025. Chart →
Independent measureChanged Mar 2026Epoch AI data →
06Safety & policy#8 of 9
How reliable is it, and how is it governed?
89%
hallucination rate of its most reliable model on AA-Omniscience, GPT-5.4 (xhigh)
xAI22%Anthropic26%Google50%DeepSeek82%OpenAI89%
Lower is better. Ranked among the tracked companies; the test covers 26 models. One reliability test: safety frameworks, evaluations and regulatory actions are logged as entries.
Independent measureChanged 2026View chart →

The ledger

What gets an entry →
An entry goes in only if it changes a number or a position on one of the six lenses. If it moves nothing, it is not news here. Newest first.

  1. September 2026

    1. 30 Sep 2026MoneyNewAnnualized revenue reached $50B, from $30B on 15 Jul 2026Anthropic's latest estimate at that date was $65B (31 Jul 2026). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$20B
    2. 3 Sep 2026CapabilityNewGPT-6 Astra scored 168.8 on the Epoch Capabilities Index, its best model#1 of the 10 tracked companies' best models; level with Anthropic at this precision. Epoch's 90% interval: 165.4 to 174.5.Independent measureEpoch AI, Benchmarking HubEpoch AI data →#1
    3. 3 Sep 2026OutputNewGPT-6 Astra released, trained with an estimated 9.4×10²⁶ FLOPAbout 14× its previous model in this data, GPT-5 (Aug 2025).Independent estimateEpoch AIView chart →9.4×10²⁶

  2. July 2026

    1. Jul 2026UseNew33.7% of employed US adults said they had used ChatGPT in the past weekProbability-based sample of 1,103 employed US adults, asked about work and personal use.Independent measureEpoch AI and IpsosEpoch AI data →
    2. 15 Jul 2026MoneyNewAnnualized revenue reached $30B, from $25B on 28 Feb 2026Anthropic's latest estimate at that date was $47B (15 May 2026). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$5B

  3. March 2026

    1. 31 Mar 2026MoneyNewClosed a funding round at a $852.0B post-money valuation, raising $12.0B in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$852.0B
    2. Mar 2026CapabilityNewGPT-5.4 High read 50.6% of ClockBench clocks correctly, the best of 15 modelsHumans manage about 90%. The next best model, Qwen 3-VL 235B Instruct, scored 39.4%.Independent measureClockBench LeaderboardView chart →#1
    3. Mar 2026UseNew31.0% of US adults said they had used ChatGPT in the past week#1 of the 7 services asked about that belong to tracked companies. 49.2% had used no AI service. Probability-based sample of 2,021 US adults.Independent measureEpoch AI and IpsosEpoch AI data →#1

  4. 2026 tests

    1. 2026 testSafety & policyNewIts most reliable model on AA-Omniscience, GPT-5.4 (xhigh), had a 89% hallucination rate#8 of 9 tracked companies; xAI's best was 22%. Rates across the 26 models tested ran from 22% to 94%.Independent measureArtificial AnalysisView chart →#8 of 9

  5. March 2026

    1. Mar 2026CapabilityNewStill #4 of 8 with 1,481; the top four labs sat within 22 Arena pointsAnthropic 1,503, xAI 1,495, Google 1,494. In May 2023 the leader was 205 points clear.Independent measureArenaView chart →#4

  6. February 2026

    1. 28 Feb 2026MoneyNewAnnualized revenue reached $25B, from $21.4B on 31 Dec 2025Anthropic's latest estimate at that date was $14B (12 Feb 2026). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$3.6B
    2. 27 Feb 2026MoneyNewClosed a funding round at a $840.0B post-money valuation, raising $110.0B in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$840.0B

  7. Early 2026 leaderboards

    1. Early 2026CapabilityNewGPT 5.1 led CaseLaw v2 legal research at 73.4%3.5 points ahead of GPT 4.1. The other 13 of the top 15 scored between 62.1% and 68.5%.Independent measureVals.aiView chart →#1

  8. December 2025

    1. 31 Dec 2025MoneyNewAnnualized revenue reached $21.4B, from $19B on 18 Dec 2025Anthropic's latest estimate at that date was $9B (31 Dec 2025). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$2.4B

  9. End of 2025

    1. End of 2025ComputeNewHad compute equal to about 1.74 million H100s available at the end of 2025Owned or rented. Epoch's 90% range: 1.25M to 2.19M. #1 of 5 labs.Independent estimateEpoch AI, AI Chip UsersEpoch AI data →1.74M

  10. 2025, full year

    1. 2025ComputeNewReported compute spend was $16.3B in 2025, up from $5.8B$8.3B on R&D and $8B on inference. Anthropic's was $6.8B.Independent estimateEpoch AIView chart →2.8×
    2. 2025OutputNewReleased 21 notable AI models in 2025, the most of any organisationGoogle 15, Alibaba 11. Epoch AI counts a model as notable if it was state of the art, highly cited or historically significant.Independent measureEpoch AIView chart →#1
    3. 2025CapabilityNewGPT-5.2 Pro solved 31.3% of FrontierMath Tier 4, the hardest maths tierThe best result in 2024 was 0%.Independent measureEpoch AIView chart →#1
    4. 2025UseNewAgentic AI overtook ChatGPT as the most-cited AI agent skill in US AI job postingsChatGPT's share of agent skill mentions fell from 25.3% in 2024 to 16.4%; agentic AI rose from 0.7% to 18.9%.Independent measureLightcastView chart →−8.9 pts
    5. 2025UseNew14,376 US AI job postings named ChatGPT as a skill in 2025Agentic AI (16,541) and AI agents (15,217) were named more often.Independent measureLightcastView chart →

  11. December 2025

    1. 18 Dec 2025MoneyNewAnnualized revenue reached $19B, from $13B on 1 Aug 2025Anthropic's latest estimate at that date was $7B (21 Oct 2025). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$6B
    2. Dec 2025CapabilityNewGPT-5.2 (high) scored 97% on MathArena's fresh competition problemsThe best model released a month earlier, Grok 4.1 Fast (Reasoning), scored 86.4%.Independent measureMathArena LeaderboardView chart →#1

  12. November 2025

    1. Nov 2025CapabilityNewFell to #4 of 8 on the Text Arena, 31 points behind GoogleGoogle led at 1,492. First to fourth spanned 31 points.Independent measureArenaView chart →#3 → #4

  13. October 2025

    1. 1 Oct 2025MoneyNewClosed a funding round at a $500.0B post-money valuationReported figureEpoch AI, AI CompaniesEpoch AI data →$500.0B

  14. August 2025

    1. 7 Aug 2025OutputNewGPT-5 released, trained with an estimated 6.6×10²⁵ FLOPAbout 13× its previous model in this data, gpt-oss-120b (Aug 2025).Independent estimateEpoch AIView chart →6.6×10²⁵
    2. 5 Aug 2025OutputNewgpt-oss-120b released, trained with an estimated 4.9×10²⁴ FLOPLess training compute than GPT-4.5 (3.8×10²⁶).Independent estimateEpoch AIView chart →4.9×10²⁴
    3. Aug 2025CapabilityNewFell to #3 of 8 on the Text Arena, 12 points behind GoogleGoogle led at 1,461. First to fourth spanned 30 points.Independent measureArenaView chart →#1 → #3
    4. 1 Aug 2025MoneyNewAnnualized revenue reached $13B, from $12B on 30 Jul 2025Anthropic's latest estimate at that date was $5B (29 Jul 2025). Annual recurring revenue (ARR); Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$1B
    5. 1 Aug 2025MoneyNewClosed a funding round at a $300.0B post-money valuation, raising $8.3B in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$300.0B

  15. July 2025

    1. 30 Jul 2025MoneyNewAnnualized revenue reached $12B, from $10B on 9 Jun 2025Anthropic's latest estimate at that date was $5B (29 Jul 2025). Annualized run rate; Epoch rates the report “Confident”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$2B
    2. Jul 2025CapabilityNewTook first place on the Text Arena with 1,482 pointsGoogle was second at 1,462. First to fourth spanned 50 points.Independent measureArenaView chart →#2 → #1

  16. 2025 studies

    1. 2025 studyCapabilityNewGPT-5 solved 205 PlanBench tasks, level with the classical planner LAMA's 204DeepSeek R1 solved 157 and Gemini 155. LAMA led or tied in six of eight domains.Independent measureCorrêa et al.View chart →#1
    2. 2025 studySafety & policyNewGPT-4o had the best safe success rate on ResponsibleRobotBench, 0.64A robot task counts only if it is completed safely around electrical, fire and human hazards. The best open-source model scored 0.35.Independent measureZhang et al.View chart →#1 of 5

  17. 2025, full year

    1. 2025ComputeNewOne medium-length GPT-4o query uses about 1.21 Wh of energyAround four Google searches. Eight a day add up to 9.7 Wh, roughly charging two phones.Independent estimateAI IndexView chart →
    2. 2025ComputeNewGPT-4o's inference uses an estimated 1.33 to 1.58 million kilolitres of water a yearThe upper estimate exceeds the annual drinking water of 1.2 million people.Independent estimateAI IndexView chart →

  18. June 2025

    1. 9 Jun 2025MoneyNewAnnualized revenue reached $10B, from $5.5B on 31 Dec 2024Anthropic's latest estimate at that date was $3B (30 May 2025). Annual recurring revenue (ARR); Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$4.5B

  19. May 2025

    1. May 2025CapabilityNewFell to #2 of 8 on the Text Arena, 34 points behind GoogleGoogle led at 1,492. First to fourth spanned 80 points.Independent measureArenaView chart →#1 → #2

  20. March 2025

    1. 31 Mar 2025MoneyNewClosed a funding round at a $300.0B post-money valuation, raising $10.0B in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$300.0B
    2. 27 Mar 2025OutputNewGPT-4o (Mar 2025) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  21. February 2025

    1. 27 Feb 2025OutputNewGPT-4.5 released, trained with an estimated 3.8×10²⁶ FLOPAbout 18× its previous model in this data, GPT-4 (Mar 2023) (Mar 2023).Independent estimateEpoch AIView chart →3.8×10²⁶

  22. January 2025

    1. 29 Jan 2025OutputNewGPT-4o (Jan 2025) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  23. December 2024

    1. 31 Dec 2024MoneyNewAnnualized revenue reached $5.5B, from $4B on 12 Sep 2024Anthropic's latest estimate at that date was $1B (31 Dec 2024). Annual recurring revenue (ARR); Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$1.5B

  24. 2024, full year

    1. 2024ComputeNewReported compute spend was $5.8B in 2024$4B on R&D and $1.8B on inference. Anthropic's was $2.5B.Independent estimateEpoch AIView chart →

  25. November 2024

    1. 20 Nov 2024OutputNewGPT-4o (Nov 2024) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  26. October 2024

    1. 2 Oct 2024MoneyNewClosed a funding round at a $157.0B post-money valuation, raising $6.6B in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$157.0B

  27. September 2024

    1. 12 Sep 2024MoneyNewAnnualized revenue reached $4B, from $3.6B on 15 Aug 2024Anthropic's latest estimate at that date was $87M (1 Jan 2024). Annual recurring revenue (ARR); Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$400M

  28. August 2024

    1. 15 Aug 2024MoneyNewAnnualized revenue reached $3.6B, from $3.4B on 12 Jun 2024Anthropic's latest estimate at that date was $87M (1 Jan 2024). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$200M
    2. 6 Aug 2024OutputNewGPT-4o (Aug 2024) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  29. June 2024

    1. 12 Jun 2024MoneyNewAnnualized revenue reached $3.4B, from $2B on 31 Dec 2023Anthropic's latest estimate at that date was $87M (1 Jan 2024). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$1.4B

  30. May 2024

    1. 13 May 2024OutputNewGPT-4o released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  31. April 2024

    1. 9 Apr 2024OutputNewGPT-4 Turbo (Apr 2024) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →
    2. Apr 2024CapabilityNewTook first place on the Text Arena with 1,338 pointsAnthropic was second at 1,323. First to fourth spanned 91 points.Independent measureArenaView chart →#2 → #1

  32. March 2024

    1. Mar 2024CapabilityNewFell to #2 of 7 on the Text Arena, 8 points behind AnthropicAnthropic led at 1,333. First to fourth spanned 95 points.Independent measureArenaView chart →#1 → #2

  33. February 2024

    1. 16 Feb 2024MoneyNewClosed a funding round at a $80.0B post-money valuationReported figureEpoch AI, AI CompaniesEpoch AI data →$80.0B

  34. December 2023

    1. 31 Dec 2023MoneyNewAnnualized revenue reached $2B, from $1.6B on 30 Dec 2023Anthropic's latest estimate at that date was $100M (1 Oct 2023). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$400M
    2. 30 Dec 2023MoneyNewAnnualized revenue reached $1.6B, from $1.3B on 10 Oct 2023Anthropic's latest estimate at that date was $100M (1 Oct 2023). Annualized run rate; Epoch rates the report “Confident”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$300M

  35. November 2023

    1. 6 Nov 2023OutputNewGPT-4 Turbo (Nov 2023) released, marked a frontier model in Epoch AI's databaseIndependent estimateEpoch AIModel record →

  36. October 2023

    1. 10 Oct 2023MoneyNewAnnualized revenue reached $1.3B, from $1B on 29 Aug 2023Anthropic's latest estimate at that date was $100M (1 Oct 2023). Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$300M

  37. August 2023

    1. 29 Aug 2023MoneyNewAnnualized revenue reached $1B, from $200M on 1 Mar 2023Annualized run rate; Epoch rates the report “Likely”.Reported figureEpoch AI, AI CompaniesEpoch AI data →+$800M

  38. June 2023

    1. 13 Jun 2023OutputNewGPT-4 (Jun 2023) released, marked a frontier model in Epoch AI's database, trained with an estimated 2.1×10²⁵ FLOPTrained on 25,000 NVIDIA A100 SXM4 40 GB.Independent estimateEpoch AIModel record →2.1×10²⁵

  39. May 2023

    1. May 2023CapabilityNewTook first place on the Text Arena with 1,322 points, its first month in this seriesGoogle was second at 1,117.Independent measureArenaView chart →#1

  40. April 2023

    1. 26 Apr 2023MoneyNewClosed a funding round at a $27.0B post-money valuation, raising $300M in equityReported figureEpoch AI, AI CompaniesEpoch AI data →$27.0B

  41. March 2023

    1. 15 Mar 2023OutputNewGPT-4 (Mar 2023) released, trained with an estimated 2.1×10²⁵ FLOPIndependent estimateEpoch AIView chart →2.1×10²⁵
    2. 15 Mar 2023ComputeNewTraining GPT-4 drew an estimated 20 MW of powerThe most in this data is Grok 3, at 110 MW.Independent estimateEpoch AIView chart →20 MW

  42. 2022, full year

    1. 2022ComputeNewReported compute spend was $416M in 2022Independent estimateEpoch AIView chart →

OpenAI's model record

Download the CSV →
Every notable model Epoch AI has logged for OpenAI: 79 in all, 61 from 2014 to 2025 and 18 in 2026 so far. Training compute, cost and hardware are Epoch's estimates, shown only where it has one.
Notable models, 2014–202561#3 of 10 companies · Epoch AI's count, as in chart 1.1.7
Frontier models15#2 of 10 companies · marked frontier in Epoch AI's database
Open weights4%2 of 57 notable models since 2023
Largest estimated training cost$366M#2 of 10 companies · GPT-4.5 (Feb 2025), 2023 dollars
Largest training cluster100,000#2 of 8 companies · NVIDIA GB200 · GPT-6 Astra
Notable models per year* 2026 to 19 Oct 2026
2014: 1120142015: 02016: 1120162017: 112018: 1120182019: 442020: 2220202021: 552022: 7720222023: 772024: 111120242025: 21212026: 18182026*
Notable models since 2023, newest first57 models
ReleasedModelAccessTraining computeEst. costTraining hardware
29 Sep 2026GPT-6.1 SolAPI or hosted access———
22 Sep 2026GPT-6 SolAPI or hosted access———
22 Sep 2026GPT-6 LunaAPI or hosted access———
8 Sep 2026GPT-Image-2.5 SunburstAPI or hosted access———
8 Sep 2026GPT-Image-2.5 FlareAPI or hosted access———
8 Sep 2026ChatGPT Images 2.5API or hosted access———
3 Sep 2026GPT-6 Astra frontierAPI or hosted access9.4×10²⁶—100,000 NVIDIA GB200
11 Aug 2026GPT-5.6 CyberAPI or hosted access———
7 May 2026GPT-5.5 CyberUnreleased———
5 May 2026GPT-5.5 InstantAPI or hosted access———
Show all 57 models since 2023
23 Apr 2026GPT-5.5 ProAPI or hosted access———
23 Apr 2026GPT-5.5API or hosted access——NVIDIA GB200
21 Apr 2026GPT Image 2API or hosted access———
14 Apr 2026GPT-5.4-CyberAPI or hosted access———
5 Mar 2026GPT-5.4 ProAPI or hosted access———
5 Mar 2026GPT-5.4API or hosted access———
12 Feb 2026GPT-5.3-Codex-SparkAPI or hosted access———
5 Feb 2026GPT-5.3 CodexAPI or hosted access———
18 Dec 2025GPT-5.2 CodexAPI or hosted access———
11 Dec 2025GPT-5.2 ProAPI or hosted access———
11 Dec 2025GPT-5.2API or hosted access———
19 Nov 2025GPT-5.1-Codex-MaxAPI or hosted access———
13 Nov 2025GPT-5.1 InstantAPI or hosted access———
13 Nov 2025GPT-5.1API or hosted access———
12 Nov 2025GPT-5.1-CodexAPI or hosted access———
7 Oct 2025GPT-5 ProAPI or hosted access———
30 Sep 2025Sora 2.0API or hosted access———
7 Aug 2025GPT-5 nanoAPI or hosted access———
7 Aug 2025GPT-5 miniAPI or hosted access———
7 Aug 2025GPT-5API or hosted access6.6×10²⁵——
5 Aug 2025gpt-oss-120bOpen weights4.9×10²⁴—NVIDIA H100 SXM5 80GB
5 Aug 2025gpt-oss-20bOpen weights5.5×10²³—NVIDIA H100 SXM5 80GB
23 Apr 2025gpt-image-1API or hosted access———
27 Mar 2025GPT-4o (Mar 2025) frontierAPI or hosted access———
19 Mar 2025o1-proAPI or hosted access———
27 Feb 2025GPT-4.5 frontierAPI or hosted access3.8×10²⁶$366M—
31 Jan 2025o3-miniAPI or hosted access———
29 Jan 2025GPT-4o (Jan 2025) frontierAPI or hosted access———
23 Jan 2025Computer-Using Agent (CUA)API or hosted access———
20 Dec 2024o3API or hosted access———
9 Dec 2024Sora TurboAPI or hosted access———
5 Dec 2024o1API or hosted access———
20 Nov 2024GPT-4o (Nov 2024) frontierAPI or hosted access———
12 Sep 2024o1-previewAPI or hosted access———
12 Sep 2024o1-miniAPI or hosted access———
6 Aug 2024GPT-4o (Aug 2024) frontierAPI or hosted access———
18 Jul 2024GPT-4o miniAPI or hosted access———
13 May 2024GPT-4o frontierAPI or hosted access———
9 Apr 2024GPT-4 Turbo (Apr 2024) frontierAPI or hosted access———
15 Feb 2024SoraAPI or hosted access———
6 Nov 2023GPT-4 Turbo (Nov 2023) frontierAPI or hosted access———
19 Oct 2023DALL·E 3API or hosted access———
28 Sep 2023GPT-3.5 Turbo InstructAPI or hosted access———
25 Sep 2023GPT-4VAPI or hosted access———
13 Jun 2023GPT-4 (Jun 2023) frontierAPI or hosted access2.1×10²⁵—25,000 NVIDIA A100 SXM4 40 GB
13 Jun 2023GPT-3.5 TurboAPI or hosted access———
15 Mar 2023GPT-4 (Mar 2023) frontierAPI or hosted access2.1×10²⁵$37M25,000 NVIDIA A100 SXM4 40 GB
Data: Epoch AI, Data on AI Models, snapshot 19 Oct 2026, CC BY 4.0. Organisations are grouped into companies as in chart 1.1.7 (for example, Google DeepMind and Google Brain count as Google). OpenAI's models as CSV.

OpenAI beside the others

All 10 companies →
The same rows for every company. A dash means no source measures it for that company yet: gaps are shown, never filled with a guess. The small number is the rank among the companies in the full table that the source covers.
LensOpenAIAnthropicGooglexAIMetaMicrosoftDeepSeekAlibabaMistral AINvidia
CapabilityEpoch Capabilities Index, its best model · data168.8#1168.8#1166.4#3156.6#6161.8#4130.5#10155.9#7157.4#5150.0#8146.0#9
OutputNotable AI models released in 2025 · chart21#17#415#25#54#60#104#611#31#93#8
MoneyAnnualized revenue, latest report · data$50B#2Sep 2026$65B#1Jul 2026—$428M#5Sep 2025——$1B#3Sep 2026—$1B#3Sep 2026—
ComputeLargest training run with a published estimate, FLOP · chart9.4×10²⁶#1GPT-6 Astra, Sep 20263.4×10²⁵#5Claude 3.7 Sonnet, Feb 20255×10²⁵#4Gemini 1.0 Ultra, Dec 20235×10²⁶#2Grok 4, Jul 20255.2×10²⁵#3Llama 4 Behemoth (preview), Apr 20258.6×10²³#9Megatron-Turing NLG 530B, Oct 2021 (with Nvidia)9.7×10²⁴#8DeepSeek-V4-Pro, Apr 20261.5×10²⁵#7Qwen3-Max, Sep 20257.7×10²³#10Mixtral 8x7B, Dec 20231.8×10²⁵#6Nemotron-4 340B, Jun 2024
UseUS adults who used its assistant in the past week, Mar 2026 · data31.0%#13.0%#620.9%#24.5%#57.5%#410.5%#30.6%#7———
Safety & policyLowest hallucination rate of its models (lower is better) · chart89%#826%#250%#322%#187%#7—82%#489%#884%#683%#5
More measures · Epoch AI data
Text Arenabest score, Mar 2026 · data1,481#41,503#11,494#31,495#21,335#8—1,424#61,449#51,416#7—
Valuationlatest closed round · data$852.0B#2$965.0B#1—$230.0B#3——$50.0B#4—$24.0B#5—
Equity raisedin all · data$182.8B#1$139.4B#2—$37.0B#3——$7.5B#4—$6.5B#5—
Stafflatest report · data4,500#23,000#46,000#1300#63,400#3—300#6—350#5—
AI chips ownedH100-equivalents · data——5.47M#10.55M#41.95M#33.32M#2————
Compute availableH100-equivalents, end of 2025 · data1.74M#11.19M#31.58M#20.63M#51.00M#4—————
Weekly use at workemployed US adults, Jul 2026 · data33.7%#17.7%#524.2%#23.3%#69.8%#414.8%#3————
Model record · Epoch AI estimates
Notable models, 2014–2025Epoch AI's count, as in chart 1.1.761#313#7194#17#887#242#47#827#66#1030#5
Frontier modelsmarked frontier in Epoch AI's database15#23#533#13#55#33#50#80#80#84#4
Open weightsshare of its notable models since 20234%2 of 570%0 of 2214%8 of 5711%1 of 973%19 of 2633%2 of 6100%10 of 1075%24 of 3267%4 of 683%15 of 18
Largest estimated training cost2023 US dollars, among models Epoch has costed$366M#2GPT-4.5$26M#5Claude 3.5 Sonnet$31M#4Gemini 1.0 Ultra$388M#1Grok 4$53M#3Llama 3.1-405B$4M#9Megatron-Turing NLG 530B$7M#8DeepSeek-R1 (May 2025)$109K#10PLUG$14M#7Mistral Large$21M#6Nemotron-4 340B
Largest training clusterchips for one model, where Epoch records it100,000#2GPT-6 Astra—57,000#3Gemini 1.0 Ultra200,000#1Grok 432,000#4Llama 4 Behemoth (preview)4,480#6Megatron-Turing NLG 530B2,048#7DeepSeek-V3 (Mar 2025)512#8CogView—6,144#5Nemotron-4 340B
Charts on AI Adoptionthat name the company3322238404714

Every chart that names OpenAI

All charts →
33 charts across 8 hubs name OpenAI or one of its models. Each has its source and a CSV of its data.
AI Benchmarks18 charts
  1. 91.2%Gemini-3.1-Pro leads MMLU-Pro at 91.2%, and the top 15 models are within 4.2 points
  2. 93%The best GPQA Diamond score rose from 36% in 2023 to 93% in 2025, above the 81.2% expert baseline
  3. 205GPT-5 solved 205 PlanBench tasks across eight domains, level with the classical planner LAMA's 204
  4. 93Disguising PlanBench tasks cut DeepSeek R1's solved total from 157 to 93; LAMA still solved 204
  5. 31.3%The best model solved 31.3% of FrontierMath's hardest problems in 2025, up from 0% in 2024
  6. 97%GPT-5.2 (high) scored 97% on MathArena in December 2025, up from 86.4% for November's best model
  7. 91.9%Aletheia scored 91.9% on IMO-ProofBench olympiad proofs in early 2026, against 7.1% for GPT-5.1
  8. 57.6%Claude Opus 4.6 led Vibe Code Bench at 57.6% in early 2026, 26 points above the lowest of eight models
  9. 70.2%Claude Opus 4.5 led τ-bench's tool-use tasks at 70.2% in early 2026; the top seven span 7.3 points
  10. 0.86Arabic.AI's LLM-X topped HELM Arabic with a mean score of 0.86, ahead of Gemini 2.5 Flash at 0.82
  11. 1,314GPT-5.2 leads the Indic LLM Arena with 1,314, ahead of GPT-5.1 (1,298) and Gemini 3 Flash (1,288)
  12. 88.6%GPT-5 scored 99.8% on standard Slovenian but 88.6% on the Cerkno dialect; 8 of 12 models fell below 60%
  13. 22.6%The best of seven AI models read analogue clocks exactly right 22.6% of the time on ClockQA
  14. 80%GPT-o1 answered 80% of CalendarQA date questions correctly; no other model of seven passed 47%
  15. 50.6%The best AI model read 50.6% of ClockBench clocks correctly in March 2026; humans manage about 90%
  16. 15.6%On Video-MMMU's measure of learning from video, human experts gain 33.1% and the best model, GPT-4o, 15.6%
  17. 69.4%No model reached 70% on MortgageTax in early 2026; Gemini 3.1 Pro Preview led at 69.4%
  18. 73.4%GPT 5.1 led CaseLaw v2 legal research at 73.4% in early 2026, 3.5 points ahead of GPT 4.1
AI Models5 charts
  1. 38×The largest US training run used about 38 times the compute of China's largest
  2. 21OpenAI released the most notable AI models in 2025: 21, ahead of Google's 15
  3. 194Google leads with 194 notable AI models since 2014, more than twice Meta's 87
  4. 1,503Anthropic, xAI, Google and OpenAI's best models were within 22 Arena points in March 2026
  5. 9.4×10²⁶The largest AI training runs now exceed 10²⁶ FLOP, led by GPT-6 Astra at 9.4×10²⁶
AI Infrastructure3 charts
  1. 23.2 WhThe most energy-hungry 2025 models used up to 23 Wh per medium-length prompt
  2. 1.21 WhA day of eight medium GPT-4o queries uses about as much energy as charging two phones
  3. 1.58M kLGPT-4o's inference uses an estimated 1.3–1.6 million kilolitres of water a year
AI Investment2 charts
  1. $65BAnthropic's annualized revenue reached $65B by July 2026; OpenAI's reached $50B by September
  2. $16.3BOpenAI's reported compute spend nearly tripled to $16.3B in 2025; Anthropic's reached $6.8B
AI Jobs2 charts
  1. 16,541US AI job postings asking for agentic AI skills rose from 151 in 2024 to 16,541 in 2025
  2. 18.9%Agentic AI overtook ChatGPT in 2025 as the most-cited AI agent skill in US AI job postings
AI Research1 chart
  1. 0.53Ai2's Asta v0 agent scored 0.53 on AstaBench at $3.40 a problem; no other agent tested passed 0.44
AI Adoption1 chart
  1. 53%Generative AI reached 53% adoption in the US three years after ChatGPT; the PC had 20% at that point
Robots & Self-Driving Cars1 chart
  1. 0.64GPT-4o scored 0.64 on ResponsibleRobotBench's safe success rate in 2025, the best of five models tested

The other companies, same lens

AI Companies →
Anthropic168.8Epoch Capabilities Index61 ledger entriesOpen tracker →Google166.4Epoch Capabilities Index48 ledger entriesOpen tracker →xAI156.6Epoch Capabilities Index37 ledger entriesOpen tracker →Meta161.8Epoch Capabilities Index33 ledger entriesOpen tracker →Microsoft130.5Epoch Capabilities Index4 ledger entriesOpen tracker →DeepSeek155.9Epoch Capabilities Index35 ledger entriesOpen tracker →Alibaba157.4Epoch Capabilities Index33 ledger entriesOpen tracker →Mistral AI150.0Epoch Capabilities Index26 ledger entriesOpen tracker →Nvidia146.0Epoch Capabilities Index7 ledger entriesOpen tracker →

How the ledger works

  1. One lens for everyone. Six lenses, same order, same definitions, for every company. No company gets an extra lens for a product it wants to talk about.
  2. Changes, not news. An entry must move a number or a rank on a lens. Launch events, interviews and hires do not count unless they do.
  3. Evidence tier on every line. Independent measure, independent estimate or company data. A company's own claims are replaced when an independent score arrives.
  4. Gaps stay visible. If no source measures a lens for a company, the page says so.
Disclosure: AI Adoption uses AI tools in its research and editorial work, including products from some of the companies tracked here. No company has any say in its page or its ledger, and the same rules apply to every company. Editorial standards · Data sources
Copied with its source. Want the next number like this in your inbox?Get Friday's numbers