Of the 875 benchmarks labs named in 2026, 570 were named by just one lab
We counted every benchmark named in the 247 launch posts labs published in 2026, and in the 133 system cards, model cards and technical reports published with them. Each shape below is one benchmark. The more labs named it, the further right it sits, and the more releases named it, the bigger it is.
The frontier labs are the nine whose best 2026 models score highest on Epoch AI's Capabilities Index, as of November 2026: Anthropic, OpenAI, Google DeepMind, Moonshot AI, Meta, DeepSeek, xAI, Qwen and Z.ai.
The data for this view didn't load, so the page shows 2026 for every lab. Try the switches again in a moment.
- Knowledge and language
- Math and science
- Coding
- Agents and tools
- Professional work
- Vision, audio and robotics
- Safety and security
- Other
releases named it
Source: Labs’ launch posts, read by Benchmark Almanac
Which benchmarks stay on top?
We track the ten benchmarks labs cite in the most releases each quarter. Only GPQA Diamond has made the top ten in every quarter since 2024. Terminal-Bench has been first for three quarters running.
Source: Labs’ launch posts, read by Benchmark Almanac
Which benchmarks stay on top?
We track the ten benchmarks frontier labs cite in the most releases each quarter. Humanity's Last Exam made frontier labs' top ten in seven of the eight quarters since 2025 began, more than any other benchmark.
Source: Labs’ launch posts, read by Benchmark Almanac
How fast do new benchmarks catch on?
DeepSWE reached ten labs 38 days after the first one cited it, the fastest of the 26 new benchmarks that got there. Half of them took 211 days or less. We count every benchmark released since 2025 that at least five labs have cited.
Source: Labs’ launch posts, read by Benchmark Almanac
How fast do new benchmarks catch on?
DeepSWE reached all nine frontier labs 48 days after the first one cited it, the fastest of the five new benchmarks that got there. Half of them took 234 days or less. We count every benchmark released since 2025 that at least five frontier labs have cited.
Source: Labs’ launch posts, read by Benchmark Almanac
Agents and tools rose from 8% of what labs reported in 2025 to 17% in 2026
What each of the 16 labs with the most releases named in 2026, by area. Each grid has 100 shapes, one for every 1% of the benchmarks the lab named across its releases. Labs with similar mixes sit side by side. "Their own" is the share the lab made itself, alone or with others, and "Shared" is the share that at least three other labs also named that year.
- Knowledge and language
- Math and science
- Coding
- Agents and tools
- Professional work
- Vision, audio and robotics
- Safety and security
- Other
Qwen29- Their own
- 5%
- Shared
- 66%
- Per release
- 44
Cohere5- Their own
- 5%
- Shared
- 81%
- Per release
- 6
Google DeepMind29- Their own
- 24%
- Shared
- 71%
- Per release
- 8
NVIDIA10- Their own
- 6%
- Shared
- 70%
- Per release
- 29.5
Microsoft5- Their own
- 7%
- Shared
- 80%
- Per release
- 10
- Their own
- 2%
- Shared
- 77%
- Per release
- 12
DeepSeek8- Their own
- 6%
- Shared
- 82%
- Per release
- 10
- Their own
- 17%
- Shared
- 72%
- Per release
- 13.5
Meta9- Their own
- 17%
- Shared
- 77%
- Per release
- 18
- Their own
- 46%
- Shared
- 63%
- Per release
- 12
- Their own
- 23%
- Shared
- 58%
- Per release
- 39
Mistral AI6- Their own
- 6%
- Shared
- 68%
- Per release
- 6
Upstage5- Their own
- 8%
- Shared
- 75%
- Per release
- 11
- Their own
- 17%
- Shared
- 74%
- Per release
- 34
- Their own
- 0%
- Shared
- 76%
- Per release
- 11
- Their own
- 9%
- Shared
- 49%
- Per release
- 6
Source: Labs’ launch posts, read by Benchmark Almanac
Terminal-Bench was named by 32 of the 45 labs that published scores in 2026
Look up any benchmark and lab. Each dot counts the lab's releases whose launch post or documents named the benchmark, in the benchmark's area color. The columns are the 12 labs with the most releases that year, and the last one adds up every other lab. Point at a dot to see the releases and the scores they reported.
Showing the 201 benchmarks that at least 3 labs named in 2026, out of 875 in total.
| Benchmark | Google | Qwen | OpenAI | Anthropic | NVIDIA | Z.ai | Meta | DeepSeek | Liquid | xAI | Mistral | Cohere | 33 others | Labs | Releases |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1Terminal-BenchStanford + Laude Institute | 8 | 14 | 10 | 13 | 6 | 5 | 9 | 7 | 1 | 3 | 2 | 1 | 49 | 32/45 | 128 |
| 2SWE-bench VerifiedPrinceton NLP + OpenAI | 1 | 11 | 8 | 5 | 2 | 2 | 2 | 2 | 1 | 1 | 36 | 30/45 | 71 | ||
| 3tau-benchSierra Research | 7 | 12 | 5 | 2 | 8 | 3 | 2 | 2 | 1 | 1 | 1 | 24 | 30/45 | 68 | |
| 4AIMEMathematical Association of America (exam), MathArena (evaluation) | 6 | 5 | 1 | 2 | 4 | 4 | 1 | 1 | 4 | 2 | 1 | 33 | 30/45 | 64 | |
| 5GPQA DiamondNYU (Rein et al.) | 8 | 20 | 10 | 5 | 8 | 4 | 3 | 3 | 4 | 1 | 42 | 29/45 | 108 | ||
| 6Humanity's Last ExamCenter for AI Safety + Scale AI | 8 | 14 | 5 | 13 | 6 | 6 | 4 | 5 | 2 | 36 | 27/45 | 99 | |||
| 7SWE-Bench ProScale AI | 4 | 11 | 9 | 11 | 1 | 2 | 3 | 2 | 1 | 1 | 31 | 27/45 | 76 | ||
| 8SWE-bench MultilingualSWE-bench team (Princeton/Stanford) | 9 | 12 | 4 | 1 | 2 | 1 | 1 | 27 | 25/45 | 57 | |||||
| 9LiveCodeBenchUC Berkeley / MIT / Cornell | 7 | 17 | 7 | 2 | 3 | 2 | 1 | 1 | 22 | 24/45 | 62 | ||||
| 10DeepSWEDatacurve | 4 | 5 | 7 | 5 | 1 | 3 | 6 | 5 | 3 | 1 | 16 | 23/45 | 56 | ||
| 11IFBenchAllenAI | 4 | 17 | 9 | 1 | 1 | 7 | 2 | 1 | 23 | 22/45 | 65 | ||||
| 12MMLU-ProTIGER-Lab (Waterloo) | 6 | 16 | 6 | 3 | 2 | 1 | 1 | 25 | 21/45 | 60 | |||||
| 13BrowseCompOpenAI | 1 | 4 | 8 | 8 | 5 | 3 | 2 | 2 | 27 | 20/45 | 60 | ||||
| 14AutomationBenchZapier | 3 | 2 | 7 | 8 | 1 | 2 | 4 | 5 | 2 | 1 | 12 | 20/45 | 47 | ||
| 15MCP AtlasScale AI | 2 | 4 | 4 | 6 | 1 | 3 | 4 | 2 | 1 | 16 | 20/45 | 43 | |||
| 16SciCodeSciCode team | 5 | 3 | 7 | 2 | 2 | 2 | 20 | 20/45 | 41 | ||||||
| 17MMMU-ProMMMU team | 10 | 15 | 9 | 2 | 1 | 2 | 1 | 1 | 1 | 1 | 19 | 19/45 | 62 | ||
| 18GDPval-AAArtificial Analysis | 6 | 1 | 3 | 12 | 3 | 3 | 6 | 2 | 2 | 19 | 19/45 | 57 | |||
| 19HMMTHarvard-MIT Mathematics Tournament (HMMT) | 14 | 2 | 3 | 2 | 20 | 19/45 | 41 | ||||||||
| 20CharXivPrinceton NLP | 6 | 17 | 1 | 7 | 2 | 1 | 3 | 1 | 17 | 17/45 | 55 | ||||
| 21ToolathlonHKUST NLP | 1 | 6 | 7 | 7 | 5 | 2 | 5 | 13 | 15/45 | 46 | |||||
| 22AA-LCRArtificial Analysis | 11 | 7 | 1 | 1 | 1 | 18 | 15/45 | 39 | |||||||
| 23IFEvalGoogle | 5 | 12 | 1 | 6 | 1 | 10 | 14/45 | 35 | |||||||
| 24OmniDocBenchShanghai AI Lab (OpenDataLab) | 6 | 13 | 4 | 1 | 1 | 1 | 1 | 8 | 14/45 | 35 | |||||
| 25τ³-bench BankingSierra | 1 | 1 | 4 | 1 | 2 | 1 | 12 | 14/45 | 22 | ||||||
| 26SimpleQAGoogle DeepMind | 2 | 4 | 1 | 1 | 3 | 10 | 14/45 | 21 | |||||||
| 27OSWorldXLANG Lab (HKU) + collaborators | 6 | 11 | 13 | 11 | 2 | 2 | 7 | 11 | 13/45 | 63 | |||||
| 28IMO-AnswerBenchGoogle DeepMind | 8 | 2 | 3 | 3 | 15 | 13/45 | 31 | ||||||||
| 29Agents' Last ExamUC Berkeley RDI + 300+ experts | 3 | 3 | 7 | 2 | 1 | 5 | 9 | 13/45 | 30 | ||||||
| 30LongBenchTsinghua KEG / Zhipu | 11 | 3 | 1 | 2 | 9 | 13/45 | 26 | ||||||||
| 31OCRBenchYuliang Liu, Xiang Bai et al. | 12 | 2 | 1 | 1 | 1 | 9 | 13/45 | 26 | |||||||
| 32SWE AtlasScale AI | 1 | 2 | 1 | 10 | 13/45 | 14 | |||||||||
| 33Berkeley Function Calling Leaderboard (BFCL)UC Berkeley (Gorilla) | 12 | 3 | 7 | 10 | 12/45 | 32 | |||||||||
| 34DeepSearchQAGoogle DeepMind | 5 | 4 | 2 | 12 | 12/45 | 23 | |||||||||
| 35MATH-Vision (MathVision)CUHK MMLab | 6 | 13 | 2 | 12 | 11/45 | 33 | |||||||||
| 36MathVistaUCLA + Microsoft | 11 | 1 | 1 | 1 | 8 | 11/45 | 22 | ||||||||
| 37Video-MMEVideo-MME team (USTC, XMU, et al.) | 16 | 2 | 1 | 11 | 10/45 | 30 | |||||||||
| 38BabyVisionUniPat AI + xbench | 10 | 1 | 1 | 2 | 2 | 10 | 10/45 | 26 | |||||||
| 39MultiChallengeScale AI | 9 | 4 | 1 | 9 | 10/45 | 23 | |||||||||
| 40MMMUMMMU team (OSU, Waterloo and others) | 12 | 1 | 1 | 2 | 6 | 10/45 | 22 | ||||||||
| 41Vals Finance AgentVals AI | 3 | 3 | 5 | 1 | 3 | 1 | 4 | 10/45 | 20 | ||||||
| 42Artificial Analysis Intelligence IndexArtificial Analysis | 2 | 4 | 2 | 3 | 1 | 1 | 6 | 10/45 | 19 | ||||||
| 43CritPtCritPt collaboration | 2 | 2 | 3 | 1 | 7 | 10/45 | 15 | ||||||||
| 44AA-OmniscienceArtificial Analysis | 1 | 1 | 9 | 4 | 2 | 4 | 6 | 9/45 | 27 | ||||||
| 45FrontierCodeCognition | 1 | 3 | 8 | 1 | 1 | 1 | 5 | 9/45 | 20 | ||||||
| 46MMStarUSTC and Shanghai AI Laboratory | 12 | 1 | 1 | 6 | 9/45 | 20 | |||||||||
| 47GDP.pdfSurge AI | 2 | 3 | 4 | 1 | 3 | 4 | 9/45 | 17 | |||||||
| 48HallusionBenchUniversity of Maryland | 9 | 1 | 1 | 6 | 9/45 | 17 | |||||||||
| 49PostTrainBenchAISA group (ELLIS Institute Tübingen and MPI-IS) + University of Tübingen + Thoughtful Lab | 1 | 4 | 2 | 1 | 8 | 9/45 | 16 | ||||||||
| 50BLINKXingyu Fu et al. | 1 | 1 | 1 | 7 | 9/45 | 10 | |||||||||
| 51ARC-AGIARC Prize Foundation | 3 | 7 | 6 | 2 | 7 | 8/45 | 25 | ||||||||
| 52SimpleVQAMultimodal Art Projection (M-A-P) et al. | 14 | 1 | 2 | 1 | 7 | 8/45 | 25 | ||||||||
| 53ScreenSpot-ProScreenSpot-Pro authors | 11 | 1 | 3 | 2 | 2 | 3 | 8/45 | 22 | |||||||
| 54AI2DAllen Institute for AI (AI2) | 11 | 1 | 1 | 5 | 8/45 | 18 | |||||||||
| 55RefCOCO / RefCOCO+ / RefCOCOgUNC Chapel Hill (Licheng Yu et al.) | 11 | 1 | 1 | 1 | 1 | 3 | 8/45 | 18 | |||||||
| 56FrontierSWEProximal | 2 | 1 | 7 | 2 | 1 | 2 | 2 | 8/45 | 17 | ||||||
| 57JobBenchUniversity of Washington + collaborators | 4 | 5 | 6 | 8/45 | 15 | ||||||||||
| 58Claw-EvalClaw-Eval team (academic) | 5 | 1 | 2 | 6 | 8/45 | 14 | |||||||||
| 59PinchBenchKilo AI | 3 | 1 | 1 | 2 | 7 | 8/45 | 14 | ||||||||
| 60Claw-Eval Pass^3Claw-Eval team | 3 | 1 | 9 | 8/45 | 13 | ||||||||||
| 61APEX-AgentsMercor | 1 | 1 | 10 | 8/45 | 12 | ||||||||||
| 62MMLUUC Berkeley (Hendrycks et al.) | 1 | 3 | 1 | 1 | 1 | 4 | 8/45 | 11 | |||||||
| 63NL2Repo-BenchByteDance + M-A-P + 2077AI | 11 | 4 | 5 | 10 | 7/45 | 30 | |||||||||
| 64MMMLUOpenAI | 8 | 11 | 4 | 1 | 4 | 7/45 | 28 | ||||||||
| 65HealthBench ProfessionalOpenAI | 1 | 10 | 9 | 3 | 1 | 2 | 7/45 | 26 | |||||||
| 66GraphWalksOpenAI | 6 | 7 | 6 | 1 | 3 | 7/45 | 23 | ||||||||
| 67HMMT Nov 2025MathArena (ETH Zurich SRI Lab), problems by HMMT | 13 | 3 | 7 | 7/45 | 23 | ||||||||||
| 68MRCRGoogle DeepMind | 12 | 3 | 2 | 2 | 3 | 7/45 | 22 | ||||||||
| 69Video-MMMUNTU S-Lab (LMMs-Lab) | 1 | 13 | 6 | 7/45 | 20 | ||||||||||
| 70OfficeQA ProDatabricks | 1 | 10 | 1 | 4 | 7/45 | 16 | |||||||||
| 71OpenAI MRCROpenAI | 8 | 3 | 1 | 4 | 7/45 | 16 | |||||||||
| 72Harvey LABHarvey | 2 | 7 | 1 | 2 | 1 | 2 | 7/45 | 15 | |||||||
| 73Arena-HardArena (formerly LMArena) | 1 | 2 | 1 | 8 | 7/45 | 12 | |||||||||
| 74SWE-MarathonAbundant AI | 2 | 2 | 3 | 5 | 7/45 | 12 | |||||||||
| 75ProgramBench almost solvedMeta Superintelligence Labs, Stanford University and Harvard University | 1 | 1 | 2 | 1 | 3 | 7/45 | 8 | ||||||||
| 76SuperGPQAByteDance Seed + M-A-P | 16 | 2 | 2 | 5 | 6/45 | 25 | |||||||||
| 77ERQAGoogle DeepMind | 1 | 17 | 1 | 1 | 4 | 6/45 | 24 | ||||||||
| 78WMT24++Google | 3 | 11 | 5 | 1 | 2 | 2 | 6/45 | 24 | |||||||
| 79HealthBenchOpenAI | 1 | 11 | 8 | 3 | 6/45 | 23 | |||||||||
| 80ZeroBenchUniversity of Cambridge and others | 11 | 2 | 3 | 6 | 6/45 | 22 | |||||||||
| 81MedXpertQA-MMTsinghua University (TsinghuaC3I) | 6 | 11 | 1 | 1 | 2 | 6/45 | 21 | ||||||||
| 82MMVUYale NLP | 10 | 1 | 6 | 6/45 | 17 | ||||||||||
| 83WideSearchByteDance Seed | 7 | 10 | 6/45 | 17 | |||||||||||
| 84CountBenchGoogle Research | 11 | 1 | 1 | 3 | 6/45 | 16 | |||||||||
| 85AA-BriefcaseArtificial Analysis | 7 | 1 | 2 | 5 | 6/45 | 15 | |||||||||
| 86ChartographySurge AI | 2 | 6 | 1 | 3 | 2 | 6/45 | 14 | ||||||||
| 87SkillsBenchBenchFlow + 105 domain experts | 7 | 1 | 6 | 6/45 | 14 | ||||||||||
| 88Terminal-Bench-ScienceStanford-led community (Terminal-Bench team + domain experts) | 2 | 3 | 5 | 3 | 6/45 | 13 | |||||||||
| 89BrowseComp-ZHResearchers from Peking University, Mindverse AI, and other groups | 4 | 1 | 4 | 6/45 | 9 | ||||||||||
| 90GSM8KOpenAI | 1 | 3 | 1 | 2 | 1 | 1 | 6/45 | 9 | |||||||
| 91HumanEvalOpenAI | 1 | 3 | 2 | 3 | 6/45 | 9 | |||||||||
| 92IMOInternational Mathematical Olympiad, hosted by the Australian Maths Trust | 1 | 1 | 2 | 4 | 6/45 | 8 | |||||||||
| 93ProgramBenchMeta Superintelligence Labs, Stanford University and Harvard University | 1 | 2 | 1 | 4 | 6/45 | 8 | |||||||||
| 94ScreenSpotOS-Atlas team (Shanghai AI Lab and others) | 1 | 1 | 1 | 4 | 6/45 | 7 | |||||||||
| 95RealWorldQAxAI | 17 | 1 | 1 | 3 | 5/45 | 22 | |||||||||
| 96Codeforces ratingCodeforces | 7 | 4 | 1 | 3 | 1 | 5/45 | 16 | ||||||||
| 97BioMysteryBenchAnthropic | 2 | 10 | 3 | 5/45 | 15 | ||||||||||
| 98HealthBench HardOpenAI | 1 | 11 | 1 | 2 | 5/45 | 15 | |||||||||
| 99VITA-BenchMeituan LongCat team | 11 | 4 | 5/45 | 15 | |||||||||||
| 100OfficeQADatabricks | 1 | 10 | 1 | 2 | 5/45 | 14 | |||||||||
| 101V*Penghao Wu and Saining Xie | 8 | 1 | 1 | 3 | 5/45 | 13 | |||||||||
| 102DRACOPerplexity | 8 | 4 | 5/45 | 12 | |||||||||||
| 103FLEURSGoogle | 5 | 4 | 1 | 2 | 5/45 | 12 | |||||||||
| 104MathArena (Apex, IMO, USAMO)ETH Zurich SRI Lab | 2 | 4 | 5 | 5/45 | 11 | ||||||||||
| 105Terminal-Bench HardArtificial Analysis | 4 | 1 | 2 | 2 | 5/45 | 9 | |||||||||
| 106WorldVQAMoonshot AI | 2 | 1 | 6 | 5/45 | 9 | ||||||||||
| 107FrontierScienceOpenAI | 8 | 5/45 | 8 | ||||||||||||
| 108InfographicVQAComputer Vision Center (UAB) and IIIT Hyderabad | 4 | 1 | 1 | 1 | 1 | 5/45 | 8 | ||||||||
| 109MATHUC Berkeley (Hendrycks et al.) | 3 | 1 | 2 | 2 | 5/45 | 8 | |||||||||
| 110WebArena-VerifiedServiceNow | 2 | 1 | 1 | 1 | 2 | 5/45 | 7 | ||||||||
| 111AdvancedIFMeta Superintelligence Labs | 6 | 5/45 | 6 | ||||||||||||
| 112ChartQAYork University and Nanyang Technological University | 1 | 1 | 1 | 2 | 5/45 | 5 | |||||||||
| 113DocVQACVC, Universitat Autònoma de Barcelona and IIIT Hyderabad | 1 | 1 | 1 | 1 | 1 | 5/45 | 5 | ||||||||
| 114LogicVistaUCLA | 1 | 1 | 3 | 5/45 | 5 | ||||||||||
| 115PaperBenchOpenAI | 1 | 4 | 5/45 | 5 | |||||||||||
| 116LVBenchZhipu AI / Tsinghua | 3 | 14 | 4 | 4/45 | 21 | ||||||||||
| 117C-EvalSJTU + Tsinghua + Edinburgh | 14 | 1 | 2 | 1 | 4/45 | 18 | |||||||||
| 118MMLU-ProXUniversity of Tokyo (Li Lab) and collaborators | 11 | 5 | 1 | 1 | 4/45 | 18 | |||||||||
| 119DynaMathUniversity of Illinois Urbana-Champaign and UC Berkeley | 11 | 4 | 4/45 | 15 | |||||||||||
| 120EmbSpatialBenchFudan University | 11 | 3 | 4/45 | 14 | |||||||||||
| 121MMBenchShanghai AI Laboratory (OpenCompass) | 11 | 1 | 1 | 1 | 4/45 | 14 | |||||||||
| 122RULERNVIDIA | 4 | 6 | 1 | 1 | 4/45 | 12 | |||||||||
| 123ProgramBench tests passedMeta Superintelligence Labs, Stanford University and Harvard University | 8 | 1 | 2 | 4/45 | 11 | ||||||||||
| 124OJBenchBeijing University of Posts and Telecommunications, Tsinghua University, and others | 6 | 4 | 4/45 | 10 | |||||||||||
| 125BBEH micro averageGoogle DeepMind | 6 | 1 | 2 | 4/45 | 9 | ||||||||||
| 126GDPvalOpenAI | 5 | 1 | 1 | 2 | 4/45 | 9 | |||||||||
| 127MATH-500OpenAI | 3 | 3 | 3 | 4/45 | 9 | ||||||||||
| 128VoiceBenchCUHK-Shenzhen | 3 | 2 | 4 | 4/45 | 9 | ||||||||||
| 129Beyond AIMEByteDance Seed | 1 | 6 | 4/45 | 7 | |||||||||||
| 130HellaSwagUniversity of Washington + AllenAI | 3 | 1 | 2 | 1 | 4/45 | 7 | |||||||||
| 131MGSMGoogle Research | 1 | 3 | 2 | 1 | 4/45 | 7 | |||||||||
| 132MMAUUniversity of Maryland | 3 | 1 | 3 | 4/45 | 7 | ||||||||||
| 133USAMOMathematical Association of America (contest), MathArena (evaluation) | 4 | 1 | 2 | 4/45 | 7 | ||||||||||
| 134We-MathBeijing University of Posts and Telecommunications | 4 | 3 | 4/45 | 7 | |||||||||||
| 135Audio MCScale AI | 1 | 1 | 1 | 3 | 4/45 | 6 | |||||||||
| 136BigCodeBenchBigCode project | 1 | 2 | 2 | 1 | 4/45 | 6 | |||||||||
| 137CL-benchTencent Hunyuan (with Fudan University authors) | 1 | 5 | 4/45 | 6 | |||||||||||
| 138LMArena (Elo)Arena (formerly LMArena) | 3 | 1 | 1 | 1 | 4/45 | 6 | |||||||||
| 139MBPPGoogle Research | 3 | 3 | 4/45 | 6 | |||||||||||
| 140ResearchRubricsScale AI | 6 | 4/45 | 6 | ||||||||||||
| 141Big Bench AudioArtificial Analysis | 1 | 2 | 2 | 4/45 | 5 | ||||||||||
| 142BIG-Bench HardSuzgun et al. (Stanford University and Google Research) | 1 | 2 | 2 | 4/45 | 5 | ||||||||||
| 143ChartQA Provis-nlp (Masry et al.) | 1 | 4 | 4/45 | 5 | |||||||||||
| 144Chinese SimpleQAAlibaba (Taobao & Tmall) | 1 | 2 | 2 | 4/45 | 5 | ||||||||||
| 145CVBenchNew York University (Cambrian-1 team) | 2 | 1 | 2 | 4/45 | 5 | ||||||||||
| 146FinSearchCompByteDance Seed | 5 | 4/45 | 5 | ||||||||||||
| 147LibriSpeechPanayotov et al. (Johns Hopkins University) | 2 | 1 | 2 | 4/45 | 5 | ||||||||||
| 148LiveCodeBench ProLiveCodeBench Pro team | 1 | 1 | 2 | 1 | 4/45 | 5 | |||||||||
| 149MMSI-BenchShanghai AI Lab (OpenRobotLab) and collaborators | 1 | 4 | 4/45 | 5 | |||||||||||
| 150PerceptionBench (Moonshot)Moonshot AI | 1 | 1 | 3 | 4/45 | 5 | ||||||||||
| 151WebDev ArenaArena (formerly LMArena) | 2 | 1 | 2 | 4/45 | 5 | ||||||||||
| 152WildClawBenchInternLM (Shanghai AI Lab) | 1 | 1 | 3 | 4/45 | 5 | ||||||||||
| 153MathVerseCUHK MMLab and Shanghai AI Laboratory | 4 | 4/45 | 4 | ||||||||||||
| 154Online-Mind2WebOhio State University (OSU NLP Group) | 1 | 1 | 2 | 4/45 | 4 | ||||||||||
| 155PutnamBenchUT Austin (Tsoukalas et al.) | 1 | 1 | 1 | 1 | 4/45 | 4 | |||||||||
| 156WebVoyagerWebVoyager team (He et al., ACL 2024) | 1 | 1 | 2 | 4/45 | 4 | ||||||||||
| 157MMLU-ReduxUniversity of Edinburgh | 16 | 1 | 1 | 3/45 | 18 | ||||||||||
| 158INCLUDEEPFL and Cohere For AI | 11 | 4 | 1 | 3/45 | 16 | ||||||||||
| 159MLVUBAAI + partners | 14 | 2 | 3/45 | 16 | |||||||||||
| 160MMLongBench-DocYubo Ma et al. | 10 | 2 | 1 | 3/45 | 13 | ||||||||||
| 161BenchCADBenchCAD team | 4 | 7 | 1 | 3/45 | 12 | ||||||||||
| 162CursorBenchCursor | 7 | 2 | 3 | 3/45 | 12 | ||||||||||
| 163SWE-bench MultimodalSWE-bench team (Princeton/Stanford) | 1 | 10 | 1 | 3/45 | 12 | ||||||||||
| 164VLMs Are BlindAuburn University and University of Alberta | 8 | 2 | 3/45 | 10 | |||||||||||
| 165MCPMarkeval-sys (MCPMark team) | 5 | 4 | 3/45 | 9 | |||||||||||
| 166Multi-IFMeta | 7 | 1 | 1 | 3/45 | 9 | ||||||||||
| 167Claw-Eval-MMClaw-Eval team | 6 | 2 | 3/45 | 8 | |||||||||||
| 168MLE-benchOpenAI | 2 | 4 | 2 | 3/45 | 8 | ||||||||||
| 169Vending-BenchAndon Labs | 1 | 5 | 2 | 3/45 | 8 | ||||||||||
| 170Vision2WebZ.ai (zai-org) | 4 | 2 | 1 | 3/45 | 7 | ||||||||||
| 171AGIEvalMicrosoft | 3 | 2 | 1 | 3/45 | 6 | ||||||||||
| 172Artificial Analysis Coding Agent IndexArtificial Analysis | 4 | 1 | 1 | 3/45 | 6 | ||||||||||
| 173Seal-0Virginia Tech | 4 | 2 | 3/45 | 6 | |||||||||||
| 174Daily-OmniZiwei Zhou, Rui Wang et al. | 3 | 1 | 1 | 3/45 | 5 | ||||||||||
| 175FrontierScience-ResearchOpenAI | 2 | 1 | 2 | 3/45 | 5 | ||||||||||
| 176LongVideoBenchNTU S-Lab et al. | 5 | 3/45 | 5 | ||||||||||||
| 177MLS-BenchMLS-Bench authors (Bohan Lyu et al.) | 2 | 3 | 3/45 | 5 | |||||||||||
| 178WinoGrandeAllenAI | 3 | 1 | 1 | 3/45 | 5 | ||||||||||
| 179Bird-SQLHKU and Alibaba DAMO Academy | 1 | 2 | 1 | 3/45 | 4 | ||||||||||
| 180Full-Duplex-BenchGuan-Ting Lin, Hung-yi Lee and collaborators (National Taiwan University and others) | 1 | 2 | 1 | 3/45 | 4 | ||||||||||
| 181Global MMLU LiteCohere Labs | 1 | 3 | 3/45 | 4 | |||||||||||
| 182IMO-ProofBenchGoogle DeepMind (Superhuman Reasoning team) | 2 | 2 | 3/45 | 4 | |||||||||||
| 183SpreadsheetBenchRenmin University (RUCKBReasoning) | 2 | 2 | 3/45 | 4 | |||||||||||
| 184Vals IndexVals AI | 2 | 2 | 3/45 | 4 | |||||||||||
| 185VisuLogicShanghai AI Laboratory with several universities | 4 | 3/45 | 4 | ||||||||||||
| 186WebArenaCarnegie Mellon University | 2 | 1 | 1 | 3/45 | 4 | ||||||||||
| 187WenetSpeechNorthwestern Polytechnical University (ASLP@NPU), Mobvoi and others | 2 | 2 | 3/45 | 4 | |||||||||||
| 188$OneMillion-BenchHumanlaya + BIGAI + xbench | 1 | 2 | 3/45 | 3 | |||||||||||
| 189AMO-BenchMeituan LongCat Team | 3 | 3/45 | 3 | ||||||||||||
| 190ArtifactsBenchTencent Hunyuan | 3 | 3/45 | 3 | ||||||||||||
| 191BrowseComp-VLAlibaba Tongyi Lab | 1 | 1 | 1 | 3/45 | 3 | ||||||||||
| 192EnterpriseOps-GymServiceNow | 1 | 2 | 3/45 | 3 | |||||||||||
| 193MindCubeMLL Lab, Northwestern University, with Stanford, NYU, and University of Washington | 1 | 2 | 3/45 | 3 | |||||||||||
| 194MM-BrowseCompByteDance Seed and collaborators (Li et al.) | 1 | 2 | 3/45 | 3 | |||||||||||
| 195PhyXHui Shen et al. | 1 | 2 | 3/45 | 3 | |||||||||||
| 196PresentBenchPresentBench team | 1 | 2 | 3/45 | 3 | |||||||||||
| 197SpreadsheetBench 2Renmin University of China (RUCKBReasoning) | 3 | 3/45 | 3 | ||||||||||||
| 198SWE-Bench Pro V2 HardScale AI (co-developed with Reflection) | 3 | 3/45 | 3 | ||||||||||||
| 199TriviaQAUniversity of Washington | 1 | 1 | 1 | 3/45 | 3 | ||||||||||
| 200ViewSpatial-BenchDingming Li, Yueting Zhuang et al. | 1 | 2 | 3/45 | 3 | |||||||||||
| 201WorkSpaceBenchSJTU + ByteDance (OpenDataBox) | 1 | 2 | 3/45 | 3 |
How this is counted
A release is one launch post. When a lab put out several models under one post, that's one release. The system cards, model cards and technical reports that came with a post count toward the same release. A benchmark counts once per release, however many settings, subsets or documents report it, and with or without a score.
Versions of one test count as one benchmark in the graphics, so Terminal-Bench 2.0 and 3.0 are both Terminal-Bench. Tests with their own task sets stay apart, like SWE-bench Verified, SWE-Bench Pro and SWE-bench Multilingual. The lookup can show every version on its own.
A benchmark's color and shape come from its area, the same eight areas the rest of the almanac uses. Safety, alignment, bio and cyber evaluations are left out until you switch them on. Most come from system cards, and only a few labs publish those.
The quarterly chart counts 2024 as one column, because we have only 31 launch posts with scores from that year. The spread chart only shows benchmarks released since 2025, so the first lab we saw name one is the first lab that did.
"Frontier labs only" counts just the releases of the nine labs whose best 2026 models score highest on Epoch AI's Capabilities Index, as of November 2026. Every count on the page then compares those labs with each other, and the spread chart asks how fast a benchmark reached them.
We only count what labs reported themselves. Leaderboard and Epoch AI scores don't count here. Every number links back to its source on the benchmark's page and the model's page, and the Model cards page lists every post we read.