Skip to Sidebar Skip to Content
Company tracker 5 of 10 · 10 companies, one lens

Meta

Meta, maker of the Llama models, tracked through the same six lenses as every company on AI Adoption.
Its best model, Muse Wonder, scores 161.8 on the Epoch Capabilities Index, #4 of 10, 7.0 behind Anthropic. In March 2026, 7.5% of US adults said they had used Meta AI in the past week. It released 4 notable models in 2025, #6 of 10 by count. It owns AI chips equal to about 1.95 million H100s.
Follow MetaCompare with the others
33 ledger entries4 charts name MetaLatest entry Oct 2026Checked 20 Oct 2026

Meta 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#4 of 10
How good are its best models?
161.8
Epoch Capabilities Index score of its best model, Muse Wonder (Oct 2026)
Anthropic168.8OpenAI168.8Google166.4Meta161.8
Epoch combines many benchmarks into one scale. 90% interval for this model: 159.4 to 165.4.
Text Arena, Mar 2026: 1,335, #8 of 8. Arena chart →
Independent measureChanged Oct 2026Epoch AI data →
02Output#6 of 10
How much does it ship?
4
Notable AI models released in 2025
OpenAI21Google15Alibaba11Anthropic7Meta4
87 notable models from 2014 to 2025 and 3 in 2026 so far. Model record →
Independent measureChanged 2025View chart →
03Money
How much does it earn?
Not measured yet
No AI revenue figure for Meta in Epoch AI's company data
Epoch AI tracks revenue for AI-first companies such as OpenAI, Anthropic, Mistral AI, DeepSeek. about 3,400 staff (Jul 2025).
How a lens gets a number →
04Compute#3 of 10
How much compute and energy goes into its AI?
5.2×10²⁵
FLOP: its largest training run with a published estimate, Llama 4 Behemoth (preview) (Apr 2025)
OpenAI9.4×10²⁶xAI5×10²⁶Meta5.2×10²⁵Google5×10²⁵
Largest estimated training cost: $53M (Llama 3.1-405B), Epoch AI. Model record →
Ranked among the 10 tracked companies. Only runs with a published estimate count, so a company that discloses less can rank lower.
Owns AI chips equal to about 1.95M H100s (Mar 2026), #3 of 7 named owners; had about 1.00M H100-equivalents available at the end of 2025, #4 of 5 labs. Epoch AI data →
Independent estimateChanged Apr 2025View chart →
05Use#4 of 7
Who uses it, and for what?
7.5%
of US adults said they used Meta AI 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: 9.8%.
Independent measureChanged Mar 2026Epoch AI data →
06Safety & policy#7 of 9
How reliable is it, and how is it governed?
87%
hallucination rate of its most reliable model on AA-Omniscience, Llama 4 Maverick
xAI22%Anthropic26%Google50%DeepSeek82%Meta87%
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. October 2026

    1. 12 Oct 2026CapabilityNewMuse Wonder scored 161.8 on the Epoch Capabilities Index, its best model#4 of the 10 tracked companies' best models; Anthropic's best scored 168.8. Epoch's 90% interval: 159.4 to 165.4.Independent measureEpoch AI, Benchmarking HubEpoch AI data →#4

  2. July 2026

    1. Jul 2026UseNew9.8% of employed US adults said they had used Meta AI 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 →

  3. March 2026

    1. Mar 2026ComputeNewOwned AI chips equal to about 1.95 million Nvidia H100sEpoch's 90% range: 1.51M to 2.52M. #3 of the 7 owners Epoch lists by name.Independent estimateEpoch AI, AI Chip OwnersEpoch AI data →1.95M
    2. Mar 2026UseNew7.5% of US adults said they had used Meta AI in the past week#4 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 →#4

  4. 2026 tests

    1. 2026 testSafety & policyNewIts most reliable model on AA-Omniscience, Llama 4 Maverick, had a 87% hallucination rate#7 of 9 tracked companies; xAI's best was 22%. Rates across the 26 models tested ran from 22% to 94%.Independent measureArtificial AnalysisView chart →#7 of 9

  5. March 2026

    1. Mar 2026CapabilityNewStill #8 of 8 with 1,335; the top four labs sat within 22 Arena pointsAnthropic 1,503, xAI 1,495, Google 1,494, OpenAI 1,481. In May 2023 the leader was 205 points clear.Independent measureArenaView chart →#8

  6. End of 2025

    1. End of 2025ComputeNewHad compute equal to about 1.00 million H100s available at the end of 2025Owned or rented. Epoch's 90% range: 0.61M to 1.64M. #4 of 5 labs.Independent estimateEpoch AI, AI Chip UsersEpoch AI data →1.00M

  7. 2014–2025

    1. 2014–2025OutputNewMeta has released 87 notable AI models since 2014, second only to Google's 194OpenAI is third with 61.Independent measureEpoch AIView chart →#2

  8. 2025, full year

    1. 2025OutputNewReleased 4 notable AI models in 2025, #6 by countOpenAI 21, Google 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 →#6

  9. 2025 studies

    1. 2025 studyCapabilityNewLlama 3.2-Vision answered 11.7% of CalendarQA date questions correctlyGPT-o1 answered 80%; no other model of seven passed 47%.Independent measureSaxena et al.View chart →

  10. May 2025

    1. May 2025CapabilityNewFell to #8 of 8 on the Text Arena, 159 points behind GoogleGoogle led at 1,492. First to fourth spanned 80 points.Independent measureArenaView chart →#7 → #8

  11. April 2025

    1. 5 Apr 2025OutputNewLlama 4 Behemoth (preview) released, trained with an estimated 5.2×10²⁵ FLOPAbout 8× its previous model in this data, Llama 3.3 70B (Dec 2024).Independent estimateEpoch AIView chart →5.2×10²⁵
    2. 5 Apr 2025OutputNewLlama 4 Maverick released, trained with an estimated 2.2×10²⁴ FLOPLess training compute than Llama 4 Behemoth (preview) (5.2×10²⁵).Independent estimateEpoch AIView chart →2.2×10²⁴
    3. 5 Apr 2025ComputeNewTraining Llama 4 Behemoth (preview) drew an estimated 44 MW of powerThe most in this data is Grok 3, at 110 MW.Independent estimateEpoch AIView chart →44 MW

  12. March 2025

    1. Mar 2025CapabilityNewFell to #7 of 8 on the Text Arena, 109 points behind OpenAIOpenAI led at 1,441. First to fourth spanned 46 points.Independent measureArenaView chart →#6 → #7

  13. February 2025

    1. Feb 2025CapabilityNewFell to #6 of 8 on the Text Arena, 71 points behind OpenAIOpenAI led at 1,405. First to fourth spanned 33 points.Independent measureArenaView chart →#5 → #6

  14. 2024, full year

    1. 2024ComputeNewTraining Llama 3.1 405B emitted an estimated 8,930 tonnes of CO₂DeepSeek v3, with more parameters (671B), emitted an estimated 597 tonnes, about a fifteenth.Independent estimateAI IndexView chart →

  15. December 2024

    1. 6 Dec 2024OutputNewLlama 3.3 70B released, trained with an estimated 6.9×10²⁴ FLOPLess training compute than Llama 3.1-405B (3.8×10²⁵).Independent estimateEpoch AIView chart →6.9×10²⁴
    2. Dec 2024CapabilityNewFell to #5 of 8 on the Text Arena, 73 points behind OpenAIOpenAI led at 1,407. First to fourth spanned 56 points.Independent measureArenaView chart →#4 → #5

  16. September 2024

    1. Sep 2024CapabilityNewFell to #4 of 8 on the Text Arena, 53 points behind OpenAIOpenAI led at 1,387. First to fourth spanned 53 points.Independent measureArenaView chart →#3 → #4

  17. July 2024

    1. 23 Jul 2024OutputNewLlama 3.1-405B released, trained with an estimated 3.8×10²⁵ FLOPAbout 5× its previous model in this data, Llama 3-70B (Apr 2024).Independent estimateEpoch AIView chart →3.8×10²⁵
    2. Jul 2024CapabilityNewRose to #3 of 7 on the Text Arena, 14 points behind OpenAIOpenAI led at 1,341. First to fourth spanned 14 points.Independent measureArenaView chart →#4 → #3

  18. May 2024

    1. May 2024CapabilityNewFell to #4 of 7 on the Text Arena, 70 points behind OpenAIOpenAI led at 1,340. First to fourth spanned 70 points.Independent measureArenaView chart →#3 → #4

  19. April 2024

    1. 18 Apr 2024OutputNewLlama 3-70B released, trained with an estimated 7.9×10²⁴ FLOPAbout 10× its previous model in this data, Llama 2-70B (Jul 2023).Independent estimateEpoch AIView chart →7.9×10²⁴
    2. Apr 2024CapabilityNewRose to #3 of 7 on the Text Arena, 59 points behind OpenAIOpenAI led at 1,338. First to fourth spanned 91 points.Independent measureArenaView chart →#7 → #3

  20. March 2024

    1. Mar 2024CapabilityNewFell to #7 of 7 on the Text Arena, 181 points behind AnthropicAnthropic led at 1,333. First to fourth spanned 95 points.Independent measureArenaView chart →#6 → #7

  21. February 2024

    1. Feb 2024CapabilityNewFell to #6 of 6 on the Text Arena, 175 points behind OpenAIOpenAI led at 1,329. First to fourth spanned 113 points.Independent measureArenaView chart →#5 → #6

  22. December 2023

    1. Dec 2023CapabilityNewFell to #5 of 5 on the Text Arena, 166 points behind OpenAIOpenAI led at 1,312. First to fourth spanned 163 points.Independent measureArenaView chart →#3 → #5

  23. October 2023

    1. Oct 2023CapabilityNewFell to #3 of 5 on the Text Arena, 166 points behind OpenAIOpenAI led at 1,314. First to fourth spanned 184 points.Independent measureArenaView chart →#2 → #3

  24. August 2023

    1. Aug 2023CapabilityNewRose to #2 of 3 on the Text Arena, 146 points behind OpenAIOpenAI led at 1,313.Independent measureArenaView chart →#3 → #2

  25. July 2023

    1. 18 Jul 2023OutputNewLlama 2-70B released, trained with an estimated 8.1×10²³ FLOPIndependent estimateEpoch AIView chart →8.1×10²³

  26. May 2023

    1. May 2023CapabilityNewPlaced #3 of 3 on the Text Arena with 969 points, its first month in this seriesOpenAI led at 1,322.Independent measureArenaView chart →#3

  27. February 2023

    1. 24 Feb 2023OutputNewLLaMA-65B released, trained with an estimated 5.5×10²³ FLOPIndependent estimateEpoch AIView chart →5.5×10²³

Meta's model record

Download the CSV →
Every notable model Epoch AI has logged for Meta: 90 in all, 87 from 2014 to 2025 and 3 in 2026 so far. Training compute, cost and hardware are Epoch's estimates, shown only where it has one.
Notable models, 2014–202587#2 of 10 companies · Epoch AI's count, as in chart 1.1.7
Frontier models5#3 of 10 companies · marked frontier in Epoch AI's database
Open weights73%19 of 26 notable models since 2023
Largest estimated training cost$53M#3 of 10 companies · Llama 3.1-405B (Jul 2024), 2023 dollars
Largest training cluster32,000#4 of 8 companies · NVIDIA H100 SXM5 80GB · Llama 4 Behemoth (preview)
Notable models per year* 2026 to 19 Oct 2026
2014: 1120142015: 112016: 5520162017: 442018: 3320182019: 17172020: 7720202021: 10102022: 161620222023: 12122024: 7720242025: 442026: 332026*
Notable models since 2023, newest first26 models
ReleasedModelAccessTraining computeEst. costTraining hardware
2 Sep 2026Muse Spark 1.3API or hosted access———
5 Aug 2026Muse Spark 1.2API or hosted access———
8 Apr 2026Muse SparkAPI or hosted access———
10 Nov 2025Meta's Generative Ads Model (GEM)Unreleased———
5 Apr 2025Llama 4 ScoutOpen weights4.1×10²⁴——
5 Apr 2025Llama 4 MaverickOpen weights2.2×10²⁴——
5 Apr 2025Llama 4 Behemoth (preview) frontierUnreleased5.2×10²⁵$45M32,000 NVIDIA H100 SXM5 80GB
13 Dec 2024Apollo 7BUnreleased——128 NVIDIA A100
6 Dec 2024Llama 3.3 70BOpen weights6.9×10²⁴—NVIDIA H100 SXM5 80GB
12 Oct 2024Llama-3.1-Nemotron-70B-InstructOpen weights7.9×10²⁴——
Show all 26 models since 2023
4 Oct 2024Movie Gen VideoUnreleased1.6×10²⁴—6,144 NVIDIA H100 SXM5 80GB
24 Sep 2024Llama 3.2 11BOpen weights5.8×10²³—NVIDIA H100 SXM5 80GB
23 Jul 2024Llama 3.1-405B frontierOpen weights3.8×10²⁵$53M16,384 NVIDIA H100 SXM5 80GB
18 Apr 2024Llama 3-70BOpen weights7.9×10²⁴—NVIDIA H100 SXM5 80GB
8 Dec 2023SeamlessM4TOpen weights——NVIDIA V100
7 Dec 2023Llama GuardOpen weights1.6×10²³—NVIDIA A100 SXM4 80 GB
8 Nov 2023MultiBand DiffusionOpen weights2.6×10¹⁹$0KNVIDIA V100
18 Jul 2023Llama 2-70BOpen weights8.1×10²³$1M1,000 NVIDIA A100 SXM4 80 GB
18 Jul 2023Llama 2-7BOpen weights8.4×10²²$114KNVIDIA A100 SXM4 80 GB
8 Jun 2023MusicGenOpen weights———
9 May 2023ImageBindOpen weights——NVIDIA V100
14 Apr 2023DINOv2Open weights7.4×10²¹$10KNVIDIA A100 SXM4 40 GB
9 Apr 2023Incoder-6.7BOpen weights3×10²¹$3KNVIDIA V100
5 Apr 2023Segment Anything ModelOpen weights7.8×10²¹$16K256 NVIDIA A100
5 Mar 2023AudioGenOpen weights9.5×10²¹$9KNVIDIA A100
24 Feb 2023LLaMA-65BOpen weights5.5×10²³$578K2,048 NVIDIA A100
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). Meta's models as CSV.

Meta 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.
LensMetaOpenAIAnthropicGooglexAIMicrosoftDeepSeekAlibabaMistral AINvidia
CapabilityEpoch Capabilities Index, its best model · data161.8#4168.8#1168.8#1166.4#3156.6#6130.5#10155.9#7157.4#5150.0#8146.0#9
OutputNotable AI models released in 2025 · chart4#621#17#415#25#50#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 · chart5.2×10²⁵#3Llama 4 Behemoth (preview), Apr 20259.4×10²⁶#1GPT-6 Astra, Sep 20263.4×10²⁵#5Claude 3.7 Sonnet, Feb 20255×10²⁵#4Gemini 1.0 Ultra, Dec 20235×10²⁶#2Grok 4, Jul 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 · data7.5%#431.0%#13.0%#620.9%#24.5%#510.5%#30.6%#7———
Safety & policyLowest hallucination rate of its models (lower is better) · chart87%#789%#826%#250%#322%#1—82%#489%#884%#683%#5
More measures · Epoch AI data
Text Arenabest score, Mar 2026 · data1,335#81,481#41,503#11,494#31,495#2—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 · data3,400#34,500#23,000#46,000#1300#6—300#6—350#5—
AI chips ownedH100-equivalents · data1.95M#3——5.47M#10.55M#43.32M#2————
Compute availableH100-equivalents, end of 2025 · data1.00M#41.74M#11.19M#31.58M#20.63M#5—————
Weekly use at workemployed US adults, Jul 2026 · data9.8%#433.7%#17.7%#524.2%#23.3%#614.8%#3————
Model record · Epoch AI estimates
Notable models, 2014–2025Epoch AI's count, as in chart 1.1.787#261#313#7194#17#842#47#827#66#1030#5
Frontier modelsmarked frontier in Epoch AI's database5#315#23#533#13#53#50#80#80#84#4
Open weightsshare of its notable models since 202373%19 of 264%2 of 570%0 of 2214%8 of 5711%1 of 933%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$53M#3Llama 3.1-405B$366M#2GPT-4.5$26M#5Claude 3.5 Sonnet$31M#4Gemini 1.0 Ultra$388M#1Grok 4$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 it32,000#4Llama 4 Behemoth (preview)100,000#2GPT-6 Astra—57,000#3Gemini 1.0 Ultra200,000#1Grok 44,480#6Megatron-Turing NLG 530B2,048#7DeepSeek-V3 (Mar 2025)512#8CogView—6,144#5Nemotron-4 340B
Charts on AI Adoptionthat name the company4332223804714

Every chart that names Meta

All charts →
4 charts across 3 hubs name Meta or one of its models. Each has its source and a CSV of its data.
AI Infrastructure2 charts
  1. 110 MWTraining Grok 3 drew an estimated 110 MW, the most of any model with published data
  2. 597 tTraining DeepSeek v3 emitted a fifteenth of the CO₂ of Llama 3.1 405B, despite more parameters
AI Benchmarks1 chart
  1. 80%GPT-o1 answered 80% of CalendarQA date questions correctly; no other model of seven passed 47%
AI Models1 chart
  1. 194Google leads with 194 notable AI models since 2014, more than twice Meta's 87

The other companies, same lens

AI Companies →
OpenAI168.8Epoch Capabilities Index65 ledger entriesOpen tracker →Anthropic168.8Epoch Capabilities Index61 ledger entriesOpen tracker →Google166.4Epoch Capabilities Index48 ledger entriesOpen tracker →xAI156.6Epoch Capabilities Index37 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