Note: Overall leaderboard rankings may not reflect true model quality — individual benchmarks give a clearer picture. ARC-Challenge MMLU GPQA GSM8K Artificial Analysis Intelligence Index v4.0

Benchmarks

All benchmarks with model scores.

Benchmark Models Top Model Score
AA-Briefcase 19 Anthropic: Claude Fable 5 1587
AA-LCR 13 Xiaomi: MiMo-V2.5 Pro 73
AA-Omniscience Accuracy 14 Anthropic: Claude Fable 5 61
AA-Omniscience Index 21 Anthropic: Claude Fable 5 40
AA-Omniscience Non-Hallucination Rate 12 xAI: Grok 4.20 78
-(Artificial Analysis Word Error Rate Index: AgentTalk) 19 Google: Gemini 3.1 Pro Preview -1.5
Agent Arena - Confirmed Success (Snapshot from 20 june 2026) 17 Anthropic: Claude Fable 5 16.27
Agent Arena - Praise vs Complaint (Snapshot from 20 june 2026) 6 MiniMax: minimax-m2.7 15.42
Agents' Last Exam (Pass Rate) (Official Agent) 3 OpenAI: GPT-5.5 24
Agents' Last Exam (Last Exam) (Score) (Official Agent) 3 OpenAI: GPT-5.5 11.2
Agents' Last Exam (Near Term) (Pass Rate) (Official Agent) 2 OpenAI: GPT-5.5 38.1
AgentWorldBench (Overall) 2 Qwen: qwen-agentworld-397b-a17b 58.71
AGIEval 7 Google: Gemma 3 27B Pretrained 66.2
AI2D 11 Google: Gemma 4 31B Instruct 88.9
Aider LLM 2 OpenAI: GPT-5 88
AIME 2024 21 Mistral AI: ministral-3-14b-instruct-2512 89.8
AIME 2025 44 Google: Gemini 3 Pro Preview 100
AIME 2026 22 Z.ai: GLM 5.2 99.2
Alpaca-Eval 2 Technology Innovation Institute: falcon-h1-1.5b-instruct 28.18
AMC-23 2 Technology Innovation Institute: falcon-h1-1.5b-deep-instruct 56.56
AMO-BENCH 1 Technology Innovation Institute: falcon-h1r-7b 36.3
APEX 2 Nex AGI: Nex-N2 Pro 36.5
APEX-Agents 10 Google: Gemini 3.1 Pro Preview 33.5
APEX-Agents-AA 3 xAI: Grok 4.3 17
APEX-shortlist 1 zyphra: zaya1-8b 32.2
APEX-v1-extended 2 Z.ai: GLM 4.7 51.7
ARC-Challenge 88 Moonshot AI: Kimi K2 Instruct 97.3
ARC-AGI v2 3 OpenAI: GPT-5.4 Pro 83.3
ARC-Easy 60 Arcee AI: trinity-nano-base 94.8
Arena-Hard-Auto v2 8 Xiaomi: mimo-v2-flash 86.2
ArguAna 15 Google: EmbeddingGemma 71.54
Artificial Analysis Intelligence Index v4.0 (Maximum Reasoning) 255 Anthropic: Claude Fable 5 64.9
Artificial Analysis Intelligence Index v4.1 (Maximum Reasoning) 257 Anthropic: Claude Fable 5 60
Artificial Analysis Agentic Index (Maximum Reasoning) 33 Anthropic: Claude Fable 5 80.6
Artificial Analysis Coding Index (Maximum Reasoning) 22 Anthropic: Claude Fable 5 62
Artificial Analysis Coding Agent Index - Claude Code 7 Anthropic: Claude Fable 5 77
Artificial Analysis Openness Index 4 LLM360: k2-think-v2 16
Artificial Analysis: Speech Reasoning (Big Bench Audio) 25 Qwen: fun-realtime-audiochat 97.6
ask-my-agent bench v2 (Fallback) 6 Google: Gemma 4 31B Instruct 72.1
ask-my-agent bench v2 (Native) 2 Google: Gemma 4 31B Instruct 64
ask-my-agent bench (Fallback) 10 Google: Gemma 3 27B Instruct 91
ask-my-agent bench (Native) 15 Google: Gemma 4 E4B Instruct 93.6
AutomationBench 1 Anthropic: Claude Fable 5 17.4
AVSpecialCollision 4 NVIDIA Corporation: cosmos3-super 79.3
AVSpecialStopBehavior 4 NVIDIA Corporation: cosmos3-super 81.6
BabyVision 4 Moonshot AI: Kimi K2.6 68.5
BankerToolBench 1 MiniMax: MiniMax M3 76.1
BBEH 2 tiger-lab: general-reasoner-qwen3-4b 12.2
BBH 27 Xiaomi: MiMo-V2.5 Pro 88.4
BeyondAIME 3 Qwen: qwen3.5-397b-a17b 72.3
BFCL v3 11 JetBrains: Mellum2 Thinking 69.4
BFCL v4 11 Qwen: qwen3.7-plus 72.9
BigBenchExtraHard 6 Google: Gemma 4 12B Instruct 53
BIG-Bench Hard 9 Google: Gemma 3 27B Instruct 87.6
BigCodeBench 10 essentialai: rnj-1-instruct 57.1
BigCodeBench Hard 2 Apple: diffucoder-7b-base 12.8
Biology-Instruction 1 Intern Language Models: intern-s1-pro 52.5
BioMysteryBench Hard 1 Anthropic: Claude Fable 5 46.1
BioMysteryBench Human Solved 1 Anthropic: Claude Fable 5 83.9
BLiMP 8 SupraLabs: Supra-50M BASE 76.3
BLINK 4 OpenBMB: minicpm-v-4.6 55.5
BlinkDepth 5 NVIDIA Corporation: cosmos3-super 91.9
BlinkSpatial 5 Google: Gemma 4 31B Instruct 90.9
BlueprintBench-2 1 Anthropic: Claude Fable 5 38.6
BoolQ 18 Arcee AI: trinity-nano-base 84.3
BridgeBench BS 5 Z.ai: GLM 5.2 100
BRIGHT 1 mteb: baseline-bm25s 0.11
BrowseComp 15 OpenAI: GPT-5.5 Pro 90.1
BruMO25 2 Tencent: hunyuan-4b-instruct 62.7
BS-Bench 1 JetBrains: Mellum2 Thinking 15
Calculator Test 1 Qwen: qwen2.5-math-1.5b-instruct 76
CaseReportBench 1 LiquidAI: LFM2.5-230M 22.51
CausalVQA 4 Qwen: qwen3-vl-32b-instruct 81
CBT-CN 1 OpenAI Community: GPT-2 87.65
CBT-NE 1 OpenAI Community: GPT-2 83.4
C-Eval 10 Anthropic: claude-opus-4.5 92.2
ChartQA 4 Meta Llama: Llama 4 Scout Instruct 83.4
CharXiv (no tools) 4 Moonshot AI: Kimi K2.6 86.7
CharXiv (with tools) 1 Anthropic: Claude Opus 4.7 91
charxivr 4 sakana: fugu-ultra 86.6
CharXiv Tools 1 Anthropic: Claude Fable 5 93.2
ChemBench 1 Intern Language Models: intern-s1 83.4
ChessBench 5 xAI: Grok 4.1 Fast 58.7
CHI-Bench 3 Z.ai: GLM 5.1 18.7
Chinese-simpleQA 1 Tencent: hy3-preview-base 69.73
CIFAR-10 12 Anke Tang: clip-vit-base-patch16_cifar10 98.28
CIFAR-100 10 edadaltocg: vit_base_patch16_224_in21k_ft_cifar100 93.16
Claw-Eval 20 DeepReinforce: Ornith-1.0-397B 77.1
CL-bench 2 MiniMax: MiniMax M3 20.5
CMath 2 Tencent: hy3-preview-base 91.17
CMMLU 4 Xiaomi: MiMo-V2.5 Pro 90.2
COCOcap 3 Google: Gemma 3 27B Pretrained 116
Codeforces ELO 11 Google: Gemma 4 31B Instruct 2150
-(CoderPPL) 47 Qwen: Qwen3.5 9B -1.6914
CoLA 1 OpenAI Community: OpenAI GPT 1 45.4
Collie 4 Mistral AI: mistral-medium-3.5-128b 95.8
CommonsenseQA 6 Arcee AI: trinity-nano-base 74.9
COPA 2 OpenAI Community: OpenAI GPT 1 78.6
Cosmos-CS 4 NVIDIA Corporation: cosmos3-super 66.4
Cosmos-ER 4 NVIDIA Corporation: cosmos-reason2-32b 74.9
CountBench 1 Qwen: Qwen3.5 27B 98
CountBenchQA 7 Qwen: qwen3-vl-32b-instruct 93.6
CoVoST 3 Google: Gemma 4 12B Instruct 38.5
CowQA 33 Z.ai: GLM 5.1 97.22
CowQA-Rapid 27 Anthropic: Claude Opus 4.7 97.22
CowQA-RU 10 MiniMax: MiniMax M3 92.59
CowQA-RU-Rapid 21 Anthropic: Claude Opus 4.7 97.22
Creative Writing Benchmark 3 Anthropic: Claude Fable 5 2213
Creative Writing v3 1 zyphra: zaya1-8b 62.97
CritPt 18 OpenAI: GPT-5.5 Pro 30.6
CRUXEval 2 Technology Innovation Institute: falcon-h1-1.5b-deep-instruct 52.32
CRUXEval-I 1 Tencent: hy3-preview-base 71.19
CRUXEval-O 1 Tencent: hy3-preview-base 68.38
CRUX-I 2 Qwen: qwen3-8b-base 59.8
CRUX-O 7 essentialai: rnj-1 63.8
CTI-REALM 3 sakana: fugu-ultra 69.4
CVBench 5 NVIDIA Corporation: cosmos-reason2-32b 88.1
CyberGym 5 OpenAI: gpt-5.5-cyber 85.6
Cyber Tasks 1 OpenAI: GPT-5.5 Pro 71.4
DeepSWE 11 OpenAI: GPT-5.5 70
DocVQA 10 Qwen: qwen3-vl-32b-instruct 96
DRACO 1 MiniMax: MiniMax M3 73.2
Dr. Bench 11 Moonshot AI: Kimi K2 Thinking 67.07
DROP 21 Tencent: youtu-llm-2b 86.7
DynaMath 1 OpenBMB: minicpm-v-4.6 59.1
ECLeKTic 4 Google: Gemma 3 27B Pretrained 24.4
EcoGym - Freelance (Income) 9 Google: Gemini 3 Pro Preview 2696
EcoGym - Operation (DAU) 9 Anthropic: claude-4.5-sonnet 1595
EcoGym - Vending (Net Worth) 9 Google: Gemini 3 Pro Preview 11272
EconGym 1 OpenAI: GPT-3.5 Turbo 52900000000
EnConda-Bench 2 Tencent: youtu-llm-2b 21.5
EnterpriseOps-Gym 1 NVIDIA Corporation: nvidia-nemotron-3-ultra 33
EQBench 1 zyphra: zaya1-8b 72.95
EQ-Bench3 1 NVIDIA Corporation: NVIDIA Nemotron 3 Nano 4B 63.2
ERQA 5 NVIDIA Corporation: cosmos3-super 51.2
EuroSAT 2 Anke Tang: clip-vit-base-patch32_eurosat 99.06941652297974
EvalPlus 9 OpenAI: gpt-4o 82.4
EvasionBench 5 Z.ai: GLM 4.7 82.91
Expert-SWE 1 OpenAI: GPT-5.5 Pro 73.1
ExploitBench 2 Anthropic: Claude Fable 5 78
EyePACS 4 Google: medgemma-1.5-4b-it 76.8
FACTS Grounding 4 Google: Gemma 3 12B Instruct 75.8
FATE-H 1 Mistral AI: leanstral-1.5-119b-a6b 87
FATE-X 1 Mistral AI: leanstral-1.5-119b-a6b 34
Finance Agent v1.1 1 Anthropic: Claude Opus 4.7 64.4
Finance Agent v2 4 Google: Gemini 3.5 Flash 57.9
-(FLEURS) 3 Google: Gemma 4 12B Instruct -0.069
FloRes 4 Google: Gemma 3 27B Pretrained 48.8
FoodTruck Bench (Net Worth) 7 OpenAI: GPT-5.5 61408
FrontierCode Diamond 10 Anthropic: Claude Fable 5 29.3
FrontierCode Extended 6 Anthropic: Claude Opus 4.8 51.8
FrontierMath 3 OpenAI: GPT-5.4 Pro 50
FrontierMath Tier 1-3 3 Anthropic: Claude Fable 5 87
FrontierMath Tier 4 3 Anthropic: Claude Fable 5 87.8
FrontierScience 1 OpenAI: GPT-5.4 Pro 36.7
FrontierScience Research 1 OpenAI: GPT-5.4 Pro 36.7
FrontierSWE 8 Xiaomi: MiMo-V2.5 Pro 100
GAIA 3 Intern Language Models: intern-s1-pro 77.4
GPD.pdf 1 Anthropic: Claude Fable 5 29.8
GDPval 3 Nex AGI: Nex-N2 Pro 1585
GDPval-AA 8 Anthropic: Claude Fable 5 1932
GDPval-AA v2 8 Anthropic: claude-sonnet-5 1618
GDPval-MM 1 OpenAI: GPT-5.5 Pro 82.6
GDPval rubrics 1 MiniMax: MiniMax M3 74.8
GeneBench 1 OpenAI: GPT-5.5 Pro 33.2
Global-MMLU 1 Xiaomi: MiMo-V2.5 Pro 83.6
Global-MMLU-Lite 4 Google: Gemma 3 27B Pretrained 75.7
GLUE 1 OpenAI Community: OpenAI GPT 1 72.8
GPQA 42 Google: Gemini 3.1 Pro Preview 94.3
GPQA Diamond 77 sakana: fugu-ultra 95.5
GPQA Diamond with tools 2 OpenAI: gpt-oss-120b 80.9
GraphWalks 1 Xiaomi: MiMo-V2.5 Pro 62
GSM8K 71 Xiaomi: MiMo-V2.5 Pro 99.6
GSM-Plus 1 JetBrains: Mellum2 Thinking 87
German Traffic Sign Recognition Benchmark 2 Anke Tang: clip-vit-base-patch32_gtsrb 98.81029725074768
HallusionBench 6 OpenBMB: minicpm-v-4.6 58.1
HaluEval 1 NVIDIA Corporation: NVIDIA Nemotron 3 Nano 4B 62.2
Handwritten-Forms 3 Google: Gemini 3 Pro Preview 94.5
-(HarmBench) 1 JetBrains: Mellum2 Thinking -20.6
HealthBench 3 Moonshot AI: Kimi K2 Thinking 58
HealthBench Hard 6 Meta: Muse Spark 42.8
HealthBench Proffesional 2 Anthropic: Claude Fable 5 66
HealthSurgiBench 2 NVIDIA Corporation: cosmos-reason2-32b 53.9
HellaSwag 74 Anthropic: claude-3-opus 95.4
hmmt2025 2 Z.ai: GLM 5.2 94.4
HMMT25 6 weiboai: vibethinker-3b 95.4
HMMT Feb 25 3 Qwen: qwen3.6-27b 93.8
HMMT Feb 2026 23 Moonshot AI: Kimi K2.6 92.7
HMMT Nov 25 3 Anthropic: claude-opus-4.5 93.3
HotPotQA 7 Qwen: Qwen3 0.6B 64
HUE I2V 2 NVIDIA Corporation: cosmos3-super 89.6
HUE T2V 2 NVIDIA Corporation: cosmos3-super 89.3
HumanEval 29 Tencent: youtu-llm-2b 95.9
HumanEvalExplain(C++) 2 IBM Granite: Granite-34B-Code-Instruct-8K 42.7
HumanEvalExplain(Go) 2 IBM Granite: Granite-34B-Code-Instruct-8K 36
HumanEvalExplain(Java) 2 IBM Granite: Granite-34B-Code-Instruct-8K 50.6
HumanEvalExplain(JavaScript) 2 IBM Granite: Granite-34B-Code-Instruct-8K 45.1
HumanEvalExplain(Python) 2 IBM Granite: Granite-34B-Code-Instruct-8K 53
HumanEvalExplain(Rust) 2 IBM Granite: Granite-34B-Code-Instruct-8K 23.8
HumanEvalFix(C++) 2 IBM Granite: Granite-34B-Code-Instruct-8K 47
HumanEvalFix(Go) 2 IBM Granite: Granite-34B-Code-Instruct-8K 51.2
HumanEvalFix(Java) 2 IBM Granite: Granite-34B-Code-Instruct-8K 55.5
HumanEvalFix(JavaScript) 2 IBM Granite: Granite-34B-Code-Instruct-8K 47.6
HumanEvalFix(Python) 2 IBM Granite: Granite-34B-Code-Instruct-8K 54.9
HumanEvalFix(Rust) 2 IBM Granite: Granite-34B-Code-Instruct-8K 45.1
HumanEval+ 20 Mistral AI: mistral-small-3.2-24b-instruct-2506 92.9
HumanEvalSynthesis(C++) 2 IBM Granite: Granite-34B-Code-Instruct-8K 57.9
HumanEvalSynthesis(Go) 2 IBM Granite: Granite-34B-Code-Instruct-8K 47.6
HumanEvalSynthesis(Java) 2 IBM Granite: Granite-34B-Code-Instruct-8K 62.8
HumanEvalSynthesis(JavaScript) 2 IBM Granite: Granite-34B-Code-Instruct-8K 56.7
HumanEvalSynthesis(Python) 2 IBM Granite: Granite-34B-Code-Instruct-8K 62.2
HumanEvalSynthesis(Rust) 2 IBM Granite: Granite-20B-Code-Instruct-8K 42.7
Humanity's Last Exam (no tools) 47 NON AI - НЕ ИИ: Human baseline - Человеческий результат 90
Humanity's Last Exam (with tools) 15 Anthropic: Claude Fable 5 64.5
HWB 9 NVIDIA Corporation: cosmos3-super 71.9
IFBench 48 MiniMax: MiniMax M3 83
IFEval 35 Qwen: Qwen3.5 27B 95
ImageNet 1 ruichen0424: ab-bnn 66.89
IMDB 2 Daniel Furman: deberta-v3-base-imdb 95.77
IMO-AnswerBench 11 Z.ai: GLM 5.2 91
INCLUDE 9 Qwen: Qwen3.5 27B 82
IndicGenBench 4 Google: Gemma 3 27B Pretrained 63.4
InfoVQA 7 Qwen: qwen3-vl-32b-instruct 87.6
Instrumental Leaderboard (Arena) - Artificial Analysis (Snapshot from 17 june 2026) 16 Suno: Suno V5.5 1185
IPhO 2025 (Theory) 1 OpenAI: GPT-5.4 Pro 93.5
ITBench-AA 4 Xiaomi: MiMo-V2.5 Pro 38
KernelBench Hard 2 MiniMax: MiniMax M3 28.8
Kimi Claw 24/7 Bench 4 OpenAI: GPT-5.5 52.8
Kimi Code Bench v2 4 OpenAI: GPT-5.5 69
KingBench 3 9 Anthropic: Claude Fable 5 88.57
KnowMe-Bench T1: Extract 4 OpenAI: GPT-5 Mini 65.4
KnowMe-Bench T2: Abstain 4 OpenAI: GPT-5 Mini 71.5
KnowMe-Bench T3: Temporal 4 OpenAI: GPT-5 Mini 54.1
KnowMe-Bench T4: Order 4 OpenAI: GPT-5 Mini 47.3
KnowMe-Bench T5: Trigger 4 OpenAI: GPT-5 Mini 42.3
KnowMe-Bench T6: Mind-Body 4 DeepSeek: DeepSeek R1 41.6
KnowMe-Bench T7: Principle 4 DeepSeek: DeepSeek R1 35.3
Lambada (Accuracy) 4 OpenAI Community: GPT-2 45.99
LAMBADA (-ppl) 1 lh-tech-ai: quark-v2-0.5m -3500
LCB-Pro 25Q2 (Easy) 1 OpenBMB: minicpm5-1b 22.68
LCBv5-v6 1 Technology Innovation Institute: falcon-h1r-7b 68.6
LCB-v6 2 Intern Language Models: intern-s1-pro 74.3
Legal Agent 1 Anthropic: Claude Fable 5 13.3
LEXam Mcq 4 Choices 18 DeepSeek: DeepSeek R1 52.41
LEXam Open Question 20 DeepSeek: deepseek-v3.2-exp 56.53
lievcodebench6 1 Qwen: Qwen3 8B 41.5
LingoQA 5 NVIDIA Corporation: cosmos3-super 76.8
LiveBench 6 OpenAI: o3-mini 84.6
LiveCodeBench 6 sakana: fugu-ultra 93.2
LiveCodeBench v5 4 Mistral AI: Magistral Small 1.2 70.88
LiveCodeBench v6 29 Moonshot AI: Kimi K2.6 89.6
LiveCodeBench Pro 1 Google: Gemini 3.1 Pro Preview 96.2
LiveSQLBench 2 MiniMax: MiniMax M3 40.2
LLMArena (llmarena.ru) (Snapshot from 19 june 2026) 38 Google: Gemini 3 Pro Preview 1162
llmarena.ru - Text-to-Image (Snapshot from july 6, 2026) 7 Google: imagen-4-ultra 1066
LOCA-Bench (256k) 1 MiniMax: MiniMax M3 49.3
LogicVista 4 Google: Gemma 4 31B Instruct 57
LogiQA 1 AMD: AMD-135m 21.2
LongVideoBench 1 OpenBMB: minicpm-v-4.6 53.8
LVBench 1 OpenBMB: minicpm-v-4.6 42
ManipBench 3 Google: gemini-2.5-pro 82
MatBench 2 Intern Language Models: intern-s1 75
MATH 38 OpenAI: o3-mini 97.9
MATH-500 25 LongCat: longcat-flash-thinking 99.2
MATH-hard 2 zyphra: zaya1-base 54.15
MATH lvl5 2 Technology Innovation Institute: falcon-h1-1.5b-deep-base 24.77
MathVerse_MINI 1 OpenBMB: minicpm-v-4.6 56.1
MATH-Vision 7 Google: Gemma 4 31B Instruct 85.6
MathVista 5 Intern Language Models: intern-s1 81.5
MathVista Mini 1 OpenAI: gpt-4o 61.3
MAXIFE 1 Qwen: Qwen3.5 0.8B 39.2
MBPP 25 Qwen: Qwen3 4B 92.3
MBPP+ 22 Tencent: hy3-preview-base 78.71
MCP-Atlas 9 Anthropic: Claude Opus 4.8 81.3
MCP-Mark 1 Qwen: qwen3.7-plus 58.7
MCP Mark Verified 4 OpenAI: GPT-5.5 92.9
MDPBench 14 rednote-hilab: dots.mocr 80.5
MedMCQA 1 biomistral: biomistral-7b 48.1
MedQA 1 biomistral: biomistral-7b 50.6
MedQA 5 opts 1 biomistral: biomistral-7b 42.8
MedXpertQA 5 Qwen: Qwen3.5 27B 62
MedXPertQA MM 4 Google: Gemma 4 31B Instruct 61.3
MERA 2 T-Tech: t-pro-it-2.0 66
MGSM 9 Google: Gemma 3 27B Pretrained 74.3
MicroVQA 2 Intern Language Models: intern-s1 63.9
MiMo Coding Bench 1 Xiaomi: MiMo-V2.5 Pro 73.7
MineBench 13 OpenAI: GPT-5.5 Pro 1863
Minerva-Math 5 essentialai: rnj-1-instruct 90.8
MixEval 1 JetBrains: Mellum2 Thinking 66.9
MLS Bench Lite 4 Anthropic: Claude Opus 4.8 42.8
MLVU (M-Avg) 1 OpenBMB: minicpm-v-4.6 65.2
MMB1.1 1 OpenAI: gpt-4o 82.2
MMBC 1 Qwen: qwen3.7-plus 46.3
MMBench-Dev 4 Qwen: qwen3-vl-32b-instruct 87.8
MME 1 OpenAI: gpt-4o 2328.7
MMHal-Score 1 OpenBMB: minicpm-v-4.6 4.6
MMLU 98 OpenAI: GPT-5 92.5
MMLU-Pro 92 Anthropic: claude-opus-4.5 89.5
MMLU-ProX 1 Qwen: Qwen3.5 0.8B 34.6
MMLU-Redux 23 Anthropic: claude-opus-4.5 95.6
MMLU-STEM 11 Qwen: Qwen3 8B 92.5
MMMLU 16 Google: Gemini 3.1 Pro Preview 92.6
MMMU 12 OpenAI: gpt-5.1 85.4
MMMU-Pro 29 OpenAI: GPT-5.4 Pro 94
MMSI-Bench 5 NVIDIA Corporation: cosmos3-super 41.8
MMSIVideoBench 4 Qwen: qwen3-vl-32b-instruct 33.8
MMStar 1 Intern Language Models: intern-s1 74.9
MMVet 1 OpenAI: gpt-4o 69.1
MNIST 4 Anke Tang: clip-vit-base-patch32_mnist 99.5726227760315
MNLI Matched 1 OpenAI Community: OpenAI GPT 1 82.1
MNLI Mismatched 1 OpenAI Community: OpenAI GPT 1 81.4
Mol-Instructions 1 Intern Language Models: intern-s1-pro 48.8
MotionBench 1 OpenBMB: minicpm-v-4.6 54.2
MPRC 1 OpenAI Community: OpenAI GPT 1 82.3
MRCR v2 8 needle 4 Google: Gemma 4 31B Instruct 66.4
MS-CXR-T 4 Google: medgemma-1.5-4b-it 65.7
MSEarth-MCQ 2 Intern Language Models: intern-s1 65.7
MT-Bench 4 DeepSeek: deepseek-v2.5 90.2
MUIRBench 1 OpenBMB: minicpm-v-4.6 60.2
multichallege 1 Qwen: Qwen3.5 0.8B 18.9
MultiChallenge 1 OpenBMB: minicpm5-1b 19.48
Multi-IF 7 LiquidAI: lfm-2.5-8b-a1b 79.93
MultiPL-E 7 Qwen: qwen3-8b-base 52.4
MUSR 6 Qwen: Qwen3 4B 60.5
MVBench 4 NVIDIA Corporation: cosmos3-super 74.5
MVPBench 4 NVIDIA Corporation: cosmos3-super 70.3
Nanonets-KIE 3 Google: Gemini 3 Pro Preview 95.2
Natural Questions 4 Google: Gemma 3 27B Pretrained 36.1
NL2Repo 10 Z.ai: GLM 5.2 48.9
ObscureBench Arena vs Gemma 3 4B 2 Google: Gemma 3 4B Instruct 50
OCRBench 8 OpenBMB: minicpm-v-4.6 838
OCRBench-v2 4 NVIDIA Corporation: cosmos3-super 66.7
OCRBench_v2_CN 2 Intern Language Models: intern-s1-pro 60.6
OCRBench_v2_EN 2 Intern Language Models: intern-s1-pro 60.1
OCW 1 eleutherai: llemma_34b 11.8
OfficeQA Pro 1 MiniMax: MiniMax M3 45.1
OJBench 3 Tencent: hunyuan-4b-instruct 12.1
OKVQA 3 Google: Gemma 3 27B Pretrained 60.2
olmOCR-bench 20 infly: infinity-parser2-pro 87.6
-(OmniDocBench V1.5) 3 OpenAI: gpt-5.5-instant 87.5
-(Open ASR Leaderboard) 53 AutoArk: ark-asr-3b -4.76
OpenBookQA 12 Arcee AI: trinity-nano-base 82.2
OpenCompass 1 OpenAI: gpt-4o 69.9
Orak 1 NVIDIA Corporation: NVIDIA Nemotron 3 Nano 4B 22.9
OSWorld-Verified 13 Anthropic: Claude Fable 5 85
PAIBench-G Image2Video (Overall) 2 NVIDIA Corporation: cosmos3-super 82.8
PAIBench-G Image2Video (Domain) 2 NVIDIA Corporation: cosmos3-super 87.3
PAIBench-G Image2Video (Quality) 2 NVIDIA Corporation: cosmos3-super 78.2
PAIBench-G Text2Video (Overall) 9 NVIDIA Corporation: cosmos3-super 80
PAIBench-G Text2Video (Domain) 9 NVIDIA Corporation: cosmos3-super 86.8
PAIBench-G Text2Video (Quality) 9 NVIDIA Corporation: cosmos3-super 73.1
PaperBench 1 MiniMax: MiniMax M3 52.6
ParseBench 29 kflai: kdl-frontier-parser-nano 76.36
PathMCQA 4 Google: MedGemma 27B Instruct 71.6
PBench 6 Technology Innovation Institute: falcon-perception 57
Physics 1 Intern Language Models: intern-s1 44
PhysicsIQ I2V 2 NVIDIA Corporation: cosmos3-super 48.9
PhysicsIQ V2V 2 NVIDIA Corporation: cosmos3-super 63.4
PiBench 10 Google: Gemma 2 27B Instruct 38
PinchBench 3 Arcee AI: trinity-large-thinking 91.9
PIQA 33 Tencent: hy3-preview-base 84.39
POLLUX 8 Google: Gemma 3 27B Instruct 1.246
PolyMATH 1 Qwen: Qwen3.5 0.8B 8.2
POPE 2 tfjiangxilab: tianjiangzhuge_3b 80
PostTrainBench 5 Anthropic: Claude Opus 4.8 37.2
ProfBench (Search) 1 NVIDIA Corporation: nvidia-nemotron-3-ultra 56
Program Bench 5 OpenAI: GPT-5.5 69.1
ProteinLMBench 1 Intern Language Models: intern-s1 63.1
PTB (-ppl) 1 OpenAI Community: GPT-2 -65.85
PubMedQA 1 biomistral: biomistral-7b 77.5
PubTabNet 3 Google: Gemini 3 Pro Preview 91.4
PutnamBench 1 Mistral AI: leanstral-1.5-119b-a6b 587
QNLI 1 OpenAI Community: OpenAI GPT 1 88.1
QQP 1 OpenAI Community: OpenAI GPT 1 70.3
QwenClawBench 3 Qwen: qwen3.6-27b 53.4
QwenWebBench 3 Anthropic: claude-opus-4.5 1536
RACE 4 OpenAI Community: OpenAI GPT 1 59
RBench 7 NVIDIA Corporation: cosmos3-nano 58.4
RealWorldQA 8 Qwen: qwen3.7-plus 86.9
RedlineBench 5 OpenAI: GPT-5.5 50.5
RefCOCO 7 Intern Language Models: intern-s1-pro 91.9
RefCOCOg 1 Moondream: Moondream 3 (Preview) 88.6
RefCOCO+ 1 Moondream: Moondream 3 (Preview) 81.8
RefSpatial 4 NVIDIA Corporation: cosmos3-super 57
ReMI 3 Google: Gemma 3 27B Pretrained 44.8
ResearchClawBench 6 Z.ai: GLM 5.1 18.19
RiddleBench 6 OpenAI: gpt-oss-120b 69.26
RoboSpatialHome 4 NVIDIA Corporation: cosmos3-super 70
ROCStories 1 OpenAI Community: OpenAI GPT 1 86.5
RTE 1 OpenAI Community: OpenAI GPT 1 56
ruAIME2025 2 T-Tech: t-pro-it-2.0 64.6
RuArenaHard vs gpt-4o 2 OpenAI: gpt-4o 50
ruLCB 2 T-Tech: t-pro-it-2.0 56.3
RULER 4 NVIDIA Corporation: nvidia-nemotron-3-ultra 95
ruMMLU 2 T-Tech: t-pro-it-2.0 79
RynnBrain-Affordance 4 Google: Gemma 4 31B Instruct 87.8
RynnBrain-Area 4 Google: Gemma 4 31B Instruct 58.7
RynnBrain-Grounding 4 Qwen: qwen3-vl-32b-instruct 77
RynnBrain-Object 4 Qwen: qwen3-vl-32b-instruct 57
RynnBrain-Spatial 4 NVIDIA Corporation: cosmos-reason2-32b 42.2
RynnBrain-Trajectory 4 NVIDIA Corporation: cosmos3-super 69.3
SArena 1 Intern Language Models: intern-s1-pro 83.5
SAT 1 eleutherai: llemma_34b 71.9
SciCode 24 sakana: fugu 60.1
SciQ 16 AMD: amd-olmo-1b 93.2
SciReasoner 1 Intern Language Models: intern-s1-pro 55.5
SciTail 1 OpenAI Community: OpenAI GPT 1 88.3
ScreenSpot Desktop 1 OpenBMB: minicpm-v-4.6 51.8
ScreenSpot Mobile 1 OpenBMB: minicpm-v-4.6 71.5
ScreenSpot-Pro 45 Qwen: qwen3.7-plus 79
ScreenSpot V2 1 Intern Language Models: intern-s1-pro 93.6
ScreenSpot Web 1 OpenBMB: minicpm-v-4.6 49.5
SFE 2 Intern Language Models: intern-s1-pro 52.7
ShoppingBench 12 Qwen: qwen2.5-max 58.4
SimpleBench 5 Anthropic: Claude Fable 5 81.9
simpleQA 8 DeepSeek: deepseek-v3.2-exp 97.1
SimulBench 5 OpenAI: gpt-4 8.74
SkillsBench 3 Qwen: qwen3.6-27b 48.2
SLAVA 6 Anthropic: claude-3.5-sonnet 70.38
SmolInstruct 2 Intern Language Models: intern-s1-pro 74.8
SNLI 1 OpenAI Community: OpenAI GPT 1 89.9
SocialICA 5 Arcee AI: trinity-nano-base 93.2
SoundBench 7 ByteDance Seed: seedance-1.5-pro 7.64
SparBench 4 NVIDIA Corporation: cosmos3-super 54.9
SpatialSense VQA 3 Google: gemma-3-12b-pt 60
Speech Arena Leaderboard - Artificial Analysis (Snapshot from 18 june 2026) 28 Qwen: fun-realtime-tts 1224
SpreadSheetBench-v1 1 MiniMax: MiniMax M3 89.4
SST-2 1 OpenAI Community: OpenAI GPT 1 91.3
Stanford Cars 2 Anke Tang: clip-vit-base-patch32_stanford-cars 78.19
STS-B 1 OpenAI Community: OpenAI GPT 1 82
SuperGPQA 58 Google: Gemini 3 Pro Preview 73.75
SVG-Bench 16 OpenAI: gpt-5.2 74.4
SWAG 1 SupraLabs: Supra-50M-Reasoning 42.33
SWE Atlas-QnA 7 DeepReinforce: Ornith-1.0-397B 41.2
SWE Atlas RF 1 Nex AGI: Nex-N2 Pro 32.9
SWE Atlas-Test Writing 7 Nex AGI: Nex-N2 Pro 40
SWE-Bench 1 OpenAI: GPT-5.5 Pro 88.7
SWE-bench Multilingual 8 Anthropic: Claude Fable 5 87.3
SWE-bench Multimodal 1 Anthropic: Claude Fable 5 59
SWE-bench Pro 43 Anthropic: Claude Fable 5 80.3
SWE-bench Verified 35 Anthropic: Claude Fable 5 95
SWE-fficiency 2 MiniMax: MiniMax M3 28.8
SWE-Marathon 3 Anthropic: Claude Opus 4.7 16
TallyQA 3 Google: Gemma 3 27B Pretrained 54.3
TARBench 4 NVIDIA Corporation: cosmos3-super 48.4
TauBench V2 (Airline) 4 Arcee AI: trinity-large-thinking 88
TauBench V2 (Average) 15 Google: Gemini 3.1 Pro Preview 99.3
TauBench V2 (Retail) 4 NVIDIA Corporation: nvidia-nemotron-3-nano-30b-a3b 56.9
TauBench V2 (Telecom) 24 Anthropic: Claude Fable 5 99
TauBench V3 (Airline) 1 Mistral AI: mistral-medium-3.5-128b 72
TauBench V3 (Average) 3 Xiaomi: MiMo-V2.5 Pro 72.9
TauBench V3 (Banking) 8 Anthropic: Claude Fable 5 27
TauBench V3 (Retail) 1 Mistral AI: mistral-medium-3.5-128b 76.1
TauBench V3 (Telecom) 1 Mistral AI: mistral-medium-3.5-128b 91.4
TEDS_TEST 3 Z.ai: glm-ocr 86
Terminal-Bench 2.0 42 OpenAI: GPT-5.5 82.7
Terminal-Bench 2.1 21 OpenAI: gpt-5.6-sol-ultra 91.9
Terminal-Bench Hard 12 Anthropic: Claude Fable 5 63
Text Arena (Snapshot from 20 june 2026) 56 Anthropic: Claude Fable 5 1508
Text to Video Leaderboard (With Audio) (Arena) - Artificial Analysis (Snapshot from 18 june 2026) 15 ByteDance Seed: dreamina-seedance-2.0-720p 1218
Text-to-Video Arena (Snapshot from 20 june 2026) 4 Google: gemini-omni-flash 1527
TextVQA 3 Google: Gemma 3 27B Pretrained 68.6
TheoremQA 7 tiger-lab: general-reasoner-qwen3-4b 48.3
Toolathlon 6 OpenAI: GPT-5.5 Pro 55.6
TriviaQA 12 Google: Gemma 3 27B Pretrained 85.5
TruthfulQA 46 Yandex: YandexGPT Lite 5 73.81
Text to Image Leaderboard (Arena) - Artificial Analysis (Snapshot from 17 june 2026) 81 OpenAI: GPT Image 2 1339
Text to Image Leaderboard (Arena) - Artificial Analysis (Snapshot from 27 june 2026) 29 OpenAI: GPT Image 2 1338
TydiQA 1 Meta Llama: Llama 4 Scout Instruct 31.5
UniMERNet 3 Z.ai: glm-ocr 96.5
USAMO 2026 1 Anthropic: Claude Fable 5 97.6
US-DermMCQA 4 Google: MedGemma 27B Instruct 75.3
V* 1 Moonshot AI: Kimi K2.6 96.9
VANTAGE-2DGrounding 4 NVIDIA Corporation: cosmos3-super 76.2
VANTAGE-2DPointing 4 Google: Gemma 4 31B Instruct 76.8
VANTAGE-Astro2D 4 NVIDIA Corporation: cosmos3-super 81.5
VANTAGE-DVC 4 NVIDIA Corporation: cosmos-reason2-32b 30.1
VANTAGE-EventVerif 4 NVIDIA Corporation: cosmos-reason2-32b 73.6
VANTAGE-SOT 4 NVIDIA Corporation: cosmos3-super 62.2
VANTAGE-Temporal 4 NVIDIA Corporation: cosmos3-super 51.9
VANTAGE-VQA 4 Qwen: qwen3-vl-32b-instruct 71.3
Vending Bench 2 8 Anthropic: claude-opus-4.6 8017.59
VIBE-V2 2 MiniMax: MiniMax M3 50.1
Video-MME-v2 2 Moonshot AI: Kimi K2.5 61.1
Video-MME (w/o subs) 1 OpenBMB: minicpm-v-4.6 59.7
VideoMME (w/ sub) 1 MiniMax: MiniMax M3 85.4
Video-MMMU 1 MiniMax: MiniMax M3 84.6
VideoPhy2 5 NVIDIA Corporation: cosmos3-super 47.4
VisDrone (Accuracy) (Test set) 4 dronefreak: visdrone-yolov26x 52.91
VizWiz-VQA 2 Qwen: qwen2.5-vl-3b-instruct 60.99
VLABench Primitive Tasks - LeRobot v3.0 4 lerobot: pi0_base 44.5
VLMsAreBlind 1 Qwen: Qwen3.5 27B 97
Vocals Leaderboard (Arena) - Artificial Analysis (Snapshot from 17 june 2026) 13 Suno: Suno V5.5 1171
VQAv2 3 Google: Gemma 3 27B Pretrained 72.9
VSI-Bench 4 NVIDIA Corporation: cosmos3-super 60.9
WBench Full 3 Tencent: hunyuanvideo-1.5 74.3
WBench Navi 11 robbyant: lingbot-world-fast 78.8
WeMath 1 OpenBMB: minicpm-v-4.6 48.8
Where2Place 5 NVIDIA Corporation: cosmos3-super 71
WideSearch 2 Nex AGI: Nex-N2 Pro 75.6
WikiText103 (-ppl) 1 OpenAI Community: GPT-2 -37.5
WikiText-2 (-ppl) 56 Yandex: YandexGPT Lite 5 -7.0947
WildClawBench 13 Nex AGI: Nex-N2 Pro 53.5
WinoGrande 64 NON AI - НЕ ИИ: Human baseline - Человеческий результат 92
WMT24++ 4 Google: Gemma 3 27B Pretrained 55.7
WSC 1 AMD: AMD-135m 36.54
xbench 2 Tencent: youtu-llm-2b 19.5
XLRS-Bench 2 Intern Language Models: intern-s1 55
XQuAD 4 Google: Gemma 3 27B Pretrained 76.8
XSTest 1 JetBrains: Mellum2 Thinking 89.6
YC-Bench 16 MiniMax: MiniMax M3 2100000
ZebraLogic 4 Anthropic: claude-3.5-sonnet 33.4
ZebraLogic Easy 4 Anthropic: claude-3.5-sonnet 87.5
ZebraLogic Hard 4 Anthropic: claude-3.5-sonnet 12.4