Benchmarks
Benchmarks named in the notes, with their status and every recorded score. Each score names who ran it.
- 271benchmarks
- 1196scores
- 303independent
- 310maintainer-run
- 583vendor-reported
303independent
310maintainer-run
583vendor-reported
Category
271 of 271 benchmarks shown
Benchmarks
BenchmarkCategoryMaintainerStatusScores
AA-Briefcasev1.1composite indexAA-LCR v1.1v1.1long contextAA-Omnisciencecomposite indexAegis2.0safetyAfriMed-QAdomainAgentClinic-MedQAdomainAgentHarmsafetyAI-Needlelong contextAI2DmultimodalAIME2024/2025/2026mathAIME 20252025mathAIME242024mathAMI (IHM, SDM)speechAndroidWorldagents and tool useARC-AGI-2knowledge and reasoningARC-AGI-3knowledge and reasoningARC-cknowledge and reasoning
Arena Agenthuman preference arenaArena Alignment IndexsafetyArena Documenthuman preference arenaArena Image Editgenerative mediaArena Image-to-Videogenerative mediaArena Searchhuman preference arenaArena Text leaderboardhuman preference arenaArena Video Editgenerative mediaArena Vision leaderboardhuman preference arenaArena WebDevhuman preference arenaArena-Hardhuman preference arenaArtificial Analysis Image Arenagenerative mediaArtificial Analysis Image-to-Video leaderboardAA-Video-I2V v1.0generative mediaArtificial Analysis Intelligence Indexv4.3.2composite indexArtificial Analysis Music Arenav1.1generative mediaArtificial Analysis Speech ArenaspeechArtificial Analysis Text-to-Image ArenaAA-Image-T2I v2.0generative mediaArtificial Analysis Text-to-Speech Arenagenerative mediaArtificial Analysis Text-to-Video leaderboardgenerative mediaArtificial Analysis Video Arenagenerative media
AudioBenchspeechAutomationBenchagents and tool useAutomationBench-AAcomposite indexBenchLM long-context compositelong contextbenchlm.ai frontier compositecomposite indexBFCLagents and tool useBFCL v4agents and tool use
BrowseCompagents and tool useC-EvalChinese language
C-MAPSS turbofan degradationindustrialCALVINworld models and roboticsCASP1616domain
CausalVQAworld models and roboticsCheXpertdomainChronos Benchmark IItime seriesCMMLUChinese languageCOCO valvaldomainCodeforces ratingcodingCommon Voice 22 (CV22-76)22speech
Common Voice 6.16.1speechCommon Voice 77.0speechConvFinQAdomainCritPtknowledge and reasoning
CursorBenchcodingCWRU bearingindustrialDeepSWEv1.1codingDeepSWE (version not stated)codingDIHARD 3speechEHRQAdomainEpic-Kitchens-100 action anticipationmultimodal
Epoch Capabilities Index (ECI)composite indexFC Reward Benchagents and tool useFEP+/OpenFE binding affinity benchmarkdomain
fev-benchtime seriesFiction.LiveBenchlong context
Figure unseen-homes whole-task evaluationworld models and robotics
FinanceBenchdomainFiQA sentimentdomainFLEURSspeech
FrontierCodev1.1codingFrontierMathTiers 1-4mathFrontierMath Tier 4v2mathFrontierMath: ErdosmathFrontierSWEcodingGAIAagents and tool useGDP.pdfknowledge and reasoningGDPvalagents and tool useGDPval-AAv2.1agents and tool use
Gemini Robotics 2 vendor task evaluationsworld models and roboticsGenEvalgenerative media
Genie Sim 3.0 sim-to-real3.0world models and roboticsGEO-BenchdomainGEO-Bench-2domain
GIFT-Evaltime seriesGlobal MMLUknowledge and reasoningGlobal MMLU LiteSoutheast Asian and IndonesianGPQAknowledge and reasoningGPQA Diamondknowledge and reasoningGranite Guardian out-of-distribution safety setsafetyGraphwalkslong contextGSM8Kmath
GSMA Open-Telco LLM BenchmarksdomainHarmBenchsafetyHarvey Legal Agent BenchmarkdomainHealthBenchdomainHealthBench ConsensusdomainHealthBench HarddomainHellaSwagknowledge and reasoning
Hugging Face Open ASR LeaderboardspeechHumanEvalcodingHumanEval+codingHumanity's Last Examknowledge and reasoningHumanity's Last Exam (HLE)knowledge and reasoning
IDSV (Indonesian Speech Variability)Southeast Asian and IndonesianIFBenchknowledge and reasoningIFEvalknowledge and reasoningImageNet top-1domainIndoMMLUSoutheast Asian and IndonesianIndoNLUSoutheast Asian and IndonesianIntPhys 22world models and roboticsIOIcodingKGC-Safetysafety
Kimi Code Benchv2codingKMMLU-Proknowledge and reasoningLABBench 22domainLIBEROworld models and roboticsLIBERO-Plusworld models and roboticsLIBERO-PROworld models and roboticsLibriSpeechspeechLiveCodeBenchcodingLiveCodeBenchv6codingLiveCodeBench v5v5codingLM-AggreFactsafetyLMArena Text-to-Image Arenagenerative mediaLMArena Text-to-Video Arenagenerative mediaLongBench v2v2long contextLVISdomainMatbench DiscoverydomainMATHmathMATH500mathMBPPcodingMBPP+codingMCP Atlasagents and tool useMedAgentBenchdomainMedCalcdomainMedMCQAdomainMedQAdomain
METR task time horizonagents and tool useMGSMmathMIMIC-CXRdomain
MLCommons AILuminate Jailbreakv0.5safetyMLCommons AILuminate Safetyv1.0safetyMLPerf Edge Agentic Inferencev6.1efficiencyMLPerf Inferencev5.1efficiencyMLPerf Inferencev6.0efficiencyMLPerf Inference (Datacenter/Edge)v6.1efficiencyMLPerf Tinyv1.3efficiencyMLPerf Tinyv1.4efficiencyMLPerf Trainingv6.0efficiencyMMLongBench-Doclong contextMMLUknowledge and reasoningMMLU MeddomainMMLU-Proknowledge and reasoningMMMUmultimodalMMMU-PromultimodalMMTEBembeddingsMMTEB-R (reranking)embeddingsMolmoSpacesworld models and roboticsMRCR v2v2long contextMT-Benchknowledge and reasoningMTEB / MMTEBv4embeddingsMTEB CodeembeddingsMTEB Englishv2embeddingsMTEB multilingualembeddingsMTEB multilingualv2embeddingsMTEB-R (reranking)embeddingsMultiJailsafetyMultilingual MMLUknowledge and reasoningMVPBenchworld models and robotics
MVTec ADindustrialMVTec AD 22industrialN-CMAPSS2021industrialNAB (Numenta Anomaly Benchmark)v1.0time seriesNeuCo-benchdomainNodeGrounddomainNusaCrowdSoutheast Asian and IndonesianNusaXSoutheast Asian and Indonesian
olmOCR-BenchmultimodalOmniDocBenchv1.6multimodalOpen X-Embodiment evaluationsworld models and roboticsOpenAI MRCRv2long contextOSWorldagents and tool use
OSWorld2.0agents and tool useOSWorld-Verifiedagents and tool usePAI-Benchworld models and roboticsPan-cancer specimen-level detection (Virchow study)domainPANGAEAdomainPerceptionTestmultimodalPHM Society Data ChallengesindustrialPhysics-IQworld models and roboticsPhyWorldworld models and roboticsPolyGuardsafetyPoseBustersdomainProgramBenchcodingPRONOSTIA / FEMTO (IEEE PHM 2012)industrialProtocol QAdomainPubMedQAdomain
Qwen-Image-Benchgenerative mediaR-Benchworld models and roboticsRadGraph F1domainReal-IADindustrialReal5-OmniDocBenchmultimodal
RecIF-BenchdomainRoboArenaworld models and roboticsRoboLabworld models and roboticsRTEBembeddings
RULERlong contextSA-Co/GolddomainSA-VdomainSciCodeknowledge and reasoning
SEA-HELMSoutheast Asian and IndonesianSEA-IFEvalSoutheast Asian and IndonesianSEA-MTBenchSoutheast Asian and IndonesianSEA-SafeguardBenchsafetySeaBenchSoutheast Asian and IndonesianSeaExamSoutheast Asian and IndonesianSimplerEnvworld models and roboticsSKABtime seriesSomething-Something v2multimodalStrongREJECTsafetySWE-bench Bash OnlycodingSWE-bench FullcodingSWE-bench LitecodingSWE-bench MultilingualcodingSWE-bench Multimodalcoding
SWE-bench ProcodingSWE-bench VerifiedcodingSWE-Lancer IC DiamondcodingSWE-MarathoncodingTabArenaAll TaskstabularTALENTtabularTARworld models and roboticstau-benchagents and tool usetau2-benchagents and tool usetau3-benchagents and tool useTeleQnAdomainTempCompassmultimodalTerminal-Bench2.0agents and tool useTerminal-Bench2.1agents and tool useTerminal-Bench3.0agents and tool useTerminal-Bench4.0agents and tool useTIMEtime seriesToolathlonagents and tool useTSB-ADtime seriesTTS ArenaspeechUSMLEdomainVals Indexcomposite indexVANTAGE-Benchworld models and roboticsVBenchv1generative mediaVBench-2.02.0generative mediaVending-Bench 2agents and tool useVicuna benchmarkhuman preference arenaVideo-MMEmultimodalVisAindustrialVisual Wake Words (VWW)domainVLABenchworld models and roboticsVLGuardsafetyVoxCeleb1 (cleaned)speech
WeatherBench 22domainWebArenaagents and tool useWebArena-Verifiedagents and tool useWildGuardTestsafetyWorldScoreworld models and roboticsXSTestsafetyNo benchmark matches these filters.
Where this comes from
Taxonomy and status from the benchmark section of the research notes.