AI Hub
Leaderboards

Model rankings

A balanced intelligence index averages each model's per-category scores. Drill into a category for individual benchmarks, or sort by speed, price, and context. See what changed → How this is calculated → Embed this leaderboard →

Updated May 25, 2026 · Benchmarks via Artificial Analysis, specs & pricing via OpenRouter · Methodology · Report an error

Top intelligence
Sonar Reasoning Pro
95.7 index
Top reasoning
Claude Opus 4.7
94.2
Top math
Grok-4 Heavy
100
Fastest
Llama 3.3 70B Instruct
2220 tok/s
Cheapest
Ling-2.6-flash
$0.01/M
Longest context
Llama 4 Scout
10M
Best open-weights
DeepSeek V3.2 Speciale
89.9 index

Price vs. intelligence

Intelligence index vs. input price — up and to the left is better value.

1008468523620$0$4$8$12$16$20Input price ($/M tokens)Intelligence indexSonar Reasoning Pro: 95.7 @ $2Sonar Reasoning ProQwen3.7 Max: 92.3 @ $2.5Gemini 3.5 Flash: 92.2 @ $1.5GPT-5.3-Codex: 91.5 @ $1.75Grok 4.20 0309 v2: 91.1 @ $2Claude Opus 4.7: 90.9 @ $5Gemini 3 Flash: 90.2 @ $0.5Gemini 3 FlashGrok 4.3: 90.1 @ $1.25DeepSeek V3.2 Speciale: 89.9 @ $0.287DeepSeek V3.2 SpecialeGPT-5.2-Codex: 89.9 @ $1.75DeepSeek-V4-Flash: 89.4 @ $0.1DeepSeek-V4-FlashQwen3.5 397B A17B: 89.3 @ $0.39Qwen3.5 397B A17BGLM 4.7: 89 @ $0.4GPT-5.1: 89 @ $1.25Qwen3.6 Max: 88.8 @ $1.04Grok 4.20 0309: 88.5 @ $2GPT-5 Pro: 88.4 @ $15GPT-5.1-Codex: 88.2 @ $1.25Qwen3.6 Plus: 88.2 @ $0.325DeepSeek-V4-Pro: 88.2 @ $0.435MiMo-V2-Flash: 88 @ $0.1MiMo-V2-FlashClaude Opus 4.5: 88 @ $5Kimi K2.5: 87.9 @ $0.4GPT-5.4 mini: 87.5 @ $0.75MiniMax M2.7: 87.4 @ $0.279GPT-5 Codex: 87.1 @ $1.25DeepSeek-V3.2: 87.1 @ $0.252MiMo-V2-Pro: 87 @ $1GLM 5.1: 86.8 @ $0.98Hy3: 86.7 @ $0.066MiMo-V2.5-Pro: 86.6 @ $1GPT-5.2: 86.2 @ $1.75Grok-3 Mini: 85.9 @ $0.3Qwen3.5-27B: 85.8 @ $0.195Qwen3.5-122B-A10B: 85.7 @ $0.26Gemma 4 31B: 85.7 @ $0.12Ring-2.6-1T: 85.7 @ $0.075Kimi K2 Thinking: 85.6 @ $0.6MiMo-V2-Omni-0327: 85.5 @ $0.4KAT-Coder-Pro V2: 85.5 @ $0.3MiMo-V2.5: 84.9 @ $0.4MiniMax M2.5: 84.8 @ $0.15GPT-5.1-Codex-Mini: 84.7 @ $0.25GLM 5 Turbo: 84.7 @ $1.2o3 Pro: 84.5 @ $20Qwen3.5-35B-A3B: 84.5 @ $0.139Qwen3 235B A22B 2507: 84.2 @ $0.4Qwen3.6 27B: 84.2 @ $0.3Qwen3.6 35B A3B: 84.1 @ $0.15MiniMax M2.1: 83.6 @ $0.29DeepSeek V3.1 Terminus: 83.5 @ $0.27Gemini 3.1 Pro: 83.2 @ $2Step 3.5 Flash: 83.1 @ $0.09MiMo-V2-Omni: 82.8 @ $0.4Gemini 3 Pro: 82.8 @ $2Qwen3.5 Omni Plus: 82.6 @ $0.4Grok-3: 82.6 @ $3o1-pro: 82.5 @ $150Gemini 3.1 Flash Lite: 82.2 @ $0.25Nova 2 Lite: 82.1 @ $0.3GLM-5: 81.9 @ $0.6KAT-Coder-Pro V1: 81.8 @ $0.3GPT-5.4 nano: 81.7 @ $0.2INTELLECT-3: 81 @ $0.2Nova 2.0 Pro: 80.9 @ $1.3Grok 3 mini Reasoning: 80.9 @ $0.3GLM 5V Turbo: 80.9 @ $1.2Qwen3.5-9B: 80.6 @ $0.04GPT-5: 80.5 @ $1.25Claude Sonnet 4.5: 80.4 @ $3Qwen3-Next-80B-A3B: 80.3 @ $0.5NVIDIA Nemotron 3 Nano 30B A3B: 80.1 @ $0.1NVIDIA Nemotron 3 Super 120B A12B: 80 @ $0.3Qwen3-235B-A22B-Thinking-2507: 79.6 @ $0.3gpt-oss-120b: 79.6 @ $0.039Qwen3 Max: 79.5 @ $0.78Llama Nemotron Super 49B v1.5: 79.4 @ $0.1Claude Opus 4.6: 79.4 @ $5Gemma 4 26B A4B: 79.2 @ $0.06GPT-5 mini: 79.2 @ $0.25Qwen2.5 VL 72B Instruct: 79.1 @ $0.25Seed-OSS-36B-Instruct: 78.8 @ $0.2Grok 4 Fast: 78.7 @ $0.2Qwen3 VL 235B A22B: 78.4 @ $0.8Qwen3 VL 32B: 78.4 @ $0.7o4-mini: 78.4 @ $1.1Llama 3.1 Nemotron Ultra 253B v1: 78.3 @ $0.6Nova 2.0 Omni: 78.2 @ $0.3Grok 4: 78.2 @ $3Magistral Medium 1.2: 78.1 @ $2Nemotron Nano 9B V2: 77.6 @ $0.04Qwen3 Next 80B A3B Thinking: 77.5 @ $0.098Mercury 2: 77 @ $0.25Mistral Small 4: 76.9 @ $0.15Gemini 2.5 Pro Preview 06-05: 76.6 @ $1.25Claude Sonnet 4.6: 76.3 @ $3Qwen3 VL 30B A3B: 76.2 @ $0.2Qwen3 Max Thinking: 76.1 @ $0.78GPT-5.5: 76.1 @ $5MiniMax-M2: 76 @ $0.255Cogito v2.1: 75.8 @ $1.3MiniMax M1 80k: 75.5 @ $0.6Claude Opus 4.1: 75.4 @ $15Claude Haiku 4.5: 75.3 @ $1Trinity Large Thinking: 75.2 @ $0.22Ling-2.6-1T: 75.2 @ $0.075DeepSeek-R1: 75 @ $0.55DeepSeek VL2: 74.9 @ $9.5Qwen3 235B A22B: 74.9 @ $0.455Kimi K2.6: 74.9 @ $0.73GPT-5.4: 74.9 @ $2.5Mistral Medium 3.5: 74.8 @ $1.5Qwen3 30B A3B 2507: 74.7 @ $0.3Claude 3.7 Sonnet: 74.7 @ $3Ring-flash-2.0: 74.6 @ $0.1Claude Sonnet 4: 74.5 @ $3Qwen3.5 Omni Flash: 74.2 @ $0.1Magistral Small 1.2: 73.9 @ $0.5Qwen3 32B: 73.8 @ $0.08Qwen3 Coder Next: 73.7 @ $0.11gpt-oss-20b: 73.6 @ $0.03Kimi K2: 73.6 @ $0.57Hermes 4 - Llama-3.1 405B: 73.5 @ $1Qwen3 Omni 30B A3B: 73.4 @ $0.3Gemini 2.5 Flash: 73.1 @ $0.3GLM-4.5: 73 @ $0.6Solar Pro 3: 72.4 @ $0.15GLM-4.6: 72.4 @ $0.43Gemini 2.5 Flash-Lite: 72.3 @ $0.1Qwen3-235B-A22B-Instruct-2507: 72.2 @ $0.15DeepSeek V3.2 Exp: 72.2 @ $0.27Qwen3 30B A3B: 71.7 @ $0.09Gemini 2.5 Pro: 71.6 @ $1.25o3: 71.6 @ $2Hermes 4 - Llama-3.1 70B: 71.3 @ $0.1GPT-5 nano: 71.2 @ $0.05Kimi K2 0905: 71 @ $0.6Ministral 8B Instruct: 70.9 @ $0.1Qwen3 VL 235B A22B Instruct: 70.9 @ $0.2o1-mini: 70.5 @ $3GLM 4.5 Air: 70.4 @ $0.13Claude 3.5 Sonnet: 70.3 @ $3DeepSeek R1 Distill Llama 70B: 70.1 @ $0.1GLM 4.5V: 70.1 @ $0.6GLM 4.6V: 69.6 @ $0.3Olmo 3 32B Think: 69.5 @ $0.15NVIDIA Nemotron Nano 12B v2 VL: 69.4 @ $0.2Claude Opus 4: 69.4 @ $15Qwen3 30B A3B 2507 Instruct: 69.3 @ $0.2Qwen3 Next 80B A3B Instruct: 68.9 @ $0.09DeepSeek R1 Distill Qwen 32B: 68.4 @ $0.12ERNIE 4.5 300B A47B: 68.2 @ $0.28QwQ-32B: 67.8 @ $0.7Gemini 1.5 Pro: 67.3 @ $1.25Ling-flash-2.0: 66.9 @ $0.1Qwen3 14B: 66.8 @ $0.1Qwen3 Coder 480B A35B Instruct: 66.5 @ $0.3Kimi K2 Instruct: 66.5 @ $0.57Qwen3 VL 30B A3B Instruct: 66.5 @ $0.13Qwen3 VL 32B Instruct: 66.5 @ $0.104Pixtral-12B: 66.1 @ $0.15DeepSeek-V3 0324: 65.9 @ $0.28o1: 65.4 @ $15o3-mini: 64.1 @ $1.1Llama 4 Maverick: 63.9 @ $0.15GPT-4.1: 63.8 @ $2Qwen2.5 Max: 63.6 @ $1.6DeepSeek-V2.5: 63.4 @ $0.14DeepSeek-R1-0528: 63.3 @ $0.55QwQ-32B-Preview: 62.6 @ $0.15Grok-2: 62.4 @ $2Mistral Small 3 24B Instruct: 62.1 @ $0.1Gemini 1.5 Flash: 61.9 @ $0.15Nova Pro: 61.6 @ $0.8MiniMax-M1: 61.5 @ $0.4Llama 3.1 405B Instruct: 60.9 @ $0.89Gemini 2.0 Flash: 60.3 @ $0.1DeepSeek-V3.1: 59.8 @ $0.21GPT-4 Turbo: 59.8 @ $10GPT-4.5: 59.4 @ $75Ling-2.6-flash: 59.3 @ $0.01Qwen2.5 72B Instruct: 59.1 @ $0.36Llama 4 Scout: 58.9 @ $0.08Sonar Pro: 58.8 @ $3Mistral Medium 3: 58.6 @ $0.4Claude 3 Opus: 58.5 @ $15Gemma 3 27B: 58.4 @ $0.08Mistral Large 3: 58.3 @ $0.5GPT-4: 58.3 @ $30GPT-4.1 Mini: 58.2 @ $0.4GLM 4.7 Flash: 58.1 @ $0.06DeepSeek-V3: 58.1 @ $0.229Qwen3 8B: 57.8 @ $0.05Nova Lite: 57.7 @ $0.06Qwen3 Omni 30B A3B Instruct: 57.3 @ $0.3o1-preview: 57.3 @ $15Gemini 2.5 Flash Lite: 57 @ $0.1Sonar: 56.8 @ $1Qwen3 4B: 56.5 @ $0.1GPT-4o: 56.4 @ $2.5Reka Flash 3: 56.2 @ $0.1Llama 3.1 70B Instruct: 56 @ $0.4Command A: 55.9 @ $2.5Gemma 3 12B: 55.5 @ $0.04Qwen3 Coder 30B A3B Instruct: 55.4 @ $0.07Claude 3.5 Haiku: 54.5 @ $0.8Gemini 2.0 Flash Lite: 54.4 @ $0.075Devstral 2: 54.3 @ $0.4Llama 3.2 90B Instruct: 54 @ $0.35Nova Premier: 53.1 @ $2.5Pixtral Large: 53.1 @ $2Reka Flash: 52.9 @ $0.2Mistral Medium 3.1: 51.5 @ $0.4Mistral Small 3.2: 51 @ $0.1Mistral Small 3.1 24B Base: 50.9 @ $0.1Llama 3.3 70B Instruct: 50.6 @ $0.1Qwen2.5 Turbo: 50.3 @ $0.1Qwen2.5 Coder 32B Instruct: 50.1 @ $0.66Qwen3 VL 8B: 49.7 @ $0.2GPT-4o-mini: 49.1 @ $0.15Nova Micro: 49 @ $0.03Gemini 1.5 Flash 8B: 48.4 @ $0.07Ministral 3 14B: 47.9 @ $0.2Mistral Large 2: 47.9 @ $2Devstral Medium: 47.8 @ $0.4Phi 4: 47.6 @ $0.065LFM2-24B-A2B: 47.4 @ $0.03Qwen3 1.7B: 46.9 @ $0.1Nemotron 3 Nano Omni 30B A3B Reasoning: 46.9 @ $0.1Qwen2.5 7B Instruct: 46.6 @ $0.04Phi-4-multimodal-instruct: 46.2 @ $0.05Phi-3.5-mini-instruct: 46 @ $0.1Claude 3 Sonnet: 45.8 @ $3Gemma 3 4B: 45.8 @ $0.04Devstral Small: 45.3 @ $0.1Llama 3.1 8B Instruct: 45 @ $0.02Gemma 3 27B Instruct: 44.5 @ $0.1Llama 3.1 Nemotron 70B Instruct: 43.7 @ $1.2Mistral Small 3: 43.6 @ $0.05Ministral 3 8B: 43.3 @ $0.15Granite 4.1 8B: 43.3 @ $0.05Qwen3 VL 8B Instruct: 43 @ $0.08Jamba 1.5 Large: 42.8 @ $2Jamba 1.6 Large: 42.6 @ $2Hermes 3 - Llama-3.1 70B: 42.5 @ $0.3Mistral Small 3.1: 42.4 @ $0.1GPT-4.1 Nano: 42.1 @ $0.1Llama 3 70B Instruct: 40.8 @ $0.51Mistral Small: 40.3 @ $0.2Gemma 3 12B Instruct: 40 @ $0.1Olmo 3 7B Instruct: 40 @ $0.1Mistral Large: 39.3 @ $2Mixtral 8x22B Instruct: 39.1 @ $2Claude 3 Haiku: 38.9 @ $0.25Llama 3.2 11B Instruct: 36.7 @ $0.05Jamba Large 1.7: 36.5 @ $2Granite 4.0 H Small: 35.7 @ $0.1GPT-3.5 Turbo: 35.2 @ $0.5Gemini 1.0 Pro: 34 @ $0.5Ministral 3 3B: 33.7 @ $0.1Mistral Medium: 33.6 @ $2.8Solar Mini: 33.1 @ $0.2Phi 4 Mini Instruct: 32.6 @ $0.08Llama 3 8B Instruct: 32.4 @ $0.04Llama 3.2 3B Instruct: 30.8 @ $0.051Jamba 1.5 Mini: 29.6 @ $0.2Qwen3 0.6B: 29.3 @ $0.1Command R+: 28.9 @ $0.15Apertus 70B Instruct: 27.2 @ $0.8Mixtral 8x7B Instruct: 26.1 @ $0.5Apertus 8B Instruct: 25.6 @ $0.1Granite 4.0 Micro: 25.6 @ $0.017Jamba 1.6 Mini: 24.9 @ $0.2Gemma 3n E4B Instructed: 24.8 @ $20Mistral 7B Instruct: 14.7 @ $0.2Llama 3.2 1B Instruct: 12.1 @ $0.027Llama 2 Chat 7B: 11.3 @ $0.1

Speed vs. intelligence

Intelligence index vs. output speed — up and to the right is fast and smart.

1008468523620080160240320400Output speed (tokens/s)Intelligence indexQwen3.7 Max: 92.3 @ 203Gemini 3.5 Flash: 92.2 @ 221GPT-5.3-Codex: 91.5 @ 73Grok 4.20 0309 v2: 91.1 @ 105Claude Opus 4.7: 90.9 @ 49Gemini 3 Flash: 90.2 @ 191Grok 4.3: 90.1 @ 88GPT-5.2-Codex: 89.9 @ 106DeepSeek-V4-Flash: 89.4 @ 109Qwen3.5 397B A17B: 89.3 @ 53GLM 4.7: 89 @ 98GPT-5.1: 89 @ 115Qwen3.6 Max: 88.8 @ 36Grok 4.20 0309: 88.5 @ 97GPT-5.1-Codex: 88.2 @ 188Qwen3.6 Plus: 88.2 @ 52DeepSeek-V4-Pro: 88.2 @ 30MiMo-V2-Flash: 88 @ 145Claude Opus 4.5: 88 @ 58Kimi K2.5: 87.9 @ 35GPT-5.4 mini: 87.5 @ 162MiniMax M2.7: 87.4 @ 50GPT-5 Codex: 87.1 @ 180MiMo-V2-Pro: 87 @ 60GLM 5.1: 86.8 @ 53Hy3: 86.7 @ 100MiMo-V2.5-Pro: 86.6 @ 58GPT-5.2: 86.2 @ 73Grok-3 Mini: 85.9 @ 100Qwen3.5-27B: 85.8 @ 91Qwen3.5-122B-A10B: 85.7 @ 129Gemma 4 31B: 85.7 @ 36Ring-2.6-1T: 85.7 @ 120Kimi K2 Thinking: 85.6 @ 100MiMo-V2-Omni-0327: 85.5 @ 110KAT-Coder-Pro V2: 85.5 @ 108MiMo-V2.5: 84.9 @ 92MiniMax M2.5: 84.8 @ 87GPT-5.1-Codex-Mini: 84.7 @ 175o3 Pro: 84.5 @ 25Qwen3.5-35B-A3B: 84.5 @ 121Qwen3 235B A22B 2507: 84.2 @ 59Qwen3.6 27B: 84.2 @ 64Qwen3.6 35B A3B: 84.1 @ 169MiniMax M2.1: 83.6 @ 92Gemini 3.1 Pro: 83.2 @ 142Step 3.5 Flash: 83.1 @ 194MiMo-V2-Omni: 82.8 @ 108Gemini 3 Pro: 82.8 @ 141Qwen3.5 Omni Plus: 82.6 @ 54Step 3.5 Flash 2603: 82.6 @ 197Grok-3: 82.6 @ 100Gemini 3.1 Flash Lite: 82.2 @ 342Nova 2 Lite: 82.1 @ 229GLM-5: 81.9 @ 67KAT-Coder-Pro V1: 81.8 @ 108GPT-5.4 nano: 81.7 @ 157Nova 2.0 Pro: 80.9 @ 149Grok 3 mini Reasoning: 80.9 @ 33Qwen3.5-9B: 80.6 @ 51GPT-5: 80.5 @ 100Claude Sonnet 4.5: 80.4 @ 42Qwen3-Next-80B-A3B: 80.3 @ 147NVIDIA Nemotron 3 Nano 30B A3B: 80.1 @ 148NVIDIA Nemotron 3 Super 120B A12B: 80 @ 211gpt-oss-120b: 79.6 @ 500Qwen3 Max: 79.5 @ 45Llama Nemotron Super 49B v1.5: 79.4 @ 51Claude Opus 4.6: 79.4 @ 48Gemma 4 26B A4B: 79.2 @ 66GPT-5 mini: 79.2 @ 200Seed-OSS-36B-Instruct: 78.8 @ 37Grok 4 Fast: 78.7 @ 90Qwen3 VL 235B A22B: 78.4 @ 34Qwen3 VL 32B: 78.4 @ 93o4-mini: 78.4 @ 115Llama 3.1 Nemotron Ultra 253B v1: 78.3 @ 42Grok 4: 78.2 @ 100Magistral Medium 1.2: 78.1 @ 42Qwen3.5 4B: 77.1 @ 164Mercury 2: 77 @ 790Mercury 2Mistral Small 4: 76.9 @ 145Gemini 2.5 Pro Preview 06-05: 76.6 @ 85Claude Sonnet 4.6: 76.3 @ 75Qwen3 VL 30B A3B: 76.2 @ 122Qwen3 Max Thinking: 76.1 @ 45GPT-5.5: 76.1 @ 67MiniMax-M2: 76 @ 91Cogito v2.1: 75.8 @ 56Claude Opus 4.1: 75.4 @ 120Claude Haiku 4.5: 75.3 @ 100Trinity Large Thinking: 75.2 @ 129DeepSeek-R1: 75 @ 189DeepSeek VL2: 74.9 @ 22Qwen3 235B A22B: 74.9 @ 68Kimi K2.6: 74.9 @ 57GPT-5.4: 74.9 @ 84Mistral Medium 3.5: 74.8 @ 140Qwen3 30B A3B 2507: 74.7 @ 151Claude 3.7 Sonnet: 74.7 @ 101Claude Sonnet 4: 74.5 @ 101Qwen3.5 Omni Flash: 74.2 @ 235Magistral Small 1.2: 73.9 @ 106Sarvam 105B: 73.8 @ 128Qwen3 32B: 73.8 @ 328Qwen3 Coder Next: 73.7 @ 92gpt-oss-20b: 73.6 @ 1000gpt-oss-20bKimi K2: 73.6 @ 26Hermes 4 - Llama-3.1 405B: 73.5 @ 34Qwen3 Omni 30B A3B: 73.4 @ 102Gemini 2.5 Flash: 73.1 @ 85GLM-4.5: 73 @ 85GLM-4.6: 72.4 @ 85Qwen3-235B-A22B-Instruct-2507: 72.2 @ 63DeepSeek V3.2 Exp: 72.2 @ 100Qwen3 30B A3B: 71.7 @ 122Gemini 2.5 Pro: 71.6 @ 85o3: 71.6 @ 50Hermes 4 - Llama-3.1 70B: 71.3 @ 60GPT-5 nano: 71.2 @ 500Kimi K2 0905: 71 @ 16Ministral 8B Instruct: 70.9 @ 0Qwen3 VL 235B A22B Instruct: 70.9 @ 51o1-mini: 70.5 @ 115GLM 4.5 Air: 70.4 @ 63Claude 3.5 Sonnet: 70.3 @ 101DeepSeek R1 Distill Llama 70B: 70.1 @ 37GLM 4.5V: 70.1 @ 85GLM 4.6V: 69.6 @ 44NVIDIA Nemotron Nano 12B v2 VL: 69.4 @ 244Claude Opus 4: 69.4 @ 120Qwen3 30B A3B 2507 Instruct: 69.3 @ 122Qwen3 Next 80B A3B Instruct: 68.9 @ 161DeepSeek R1 Distill Qwen 32B: 68.4 @ 37NVIDIA Nemotron Nano 9B V2: 68.3 @ 129ERNIE 4.5 300B A47B: 68.2 @ 24QwQ-32B: 67.8 @ 31Gemini 1.5 Pro: 67.3 @ 85Ling-flash-2.0: 66.9 @ 91Qwen3 14B: 66.8 @ 62Qwen3 Coder 480B A35B Instruct: 66.5 @ 69Kimi K2 Instruct: 66.5 @ 45Qwen3 VL 30B A3B Instruct: 66.5 @ 123Qwen3 VL 32B Instruct: 66.5 @ 76Pixtral-12B: 66.1 @ 0o1: 65.4 @ 66o3-mini: 64.1 @ 115Llama 4 Maverick: 63.9 @ 639GPT-4.1: 63.8 @ 100Qwen2.5 Max: 63.6 @ 50LongCat Flash Lite: 63.6 @ 110DeepSeek-V2.5: 63.4 @ 100Sarvam 30B: 63.3 @ 214DeepSeek-R1-0528: 63.3 @ 45QwQ-32B-Preview: 62.6 @ 99Grok-2: 62.4 @ 85Mistral Small 3 24B Instruct: 62.1 @ 134Gemini 1.5 Flash: 61.9 @ 150Nova Pro: 61.6 @ 100Llama 3.1 405B Instruct: 60.9 @ 100Gemini 2.0 Flash: 60.3 @ 183GPT-4 Turbo: 59.8 @ 100GPT-4.5: 59.4 @ 50Qwen2.5 72B Instruct: 59.1 @ 100Llama 4 Scout: 58.9 @ 776Llama 4 ScoutMistral Medium 3: 58.6 @ 32Claude 3 Opus: 58.5 @ 120Gemma 3 27B: 58.4 @ 33Mistral Large 3: 58.3 @ 54GPT-4: 58.3 @ 104GPT-4.1 Mini: 58.2 @ 150GLM 4.7 Flash: 58.1 @ 113DeepSeek-V3: 58.1 @ 100Qwen3 8B: 57.8 @ 69Nova Lite: 57.7 @ 100Qwen3 Omni 30B A3B Instruct: 57.3 @ 103o1-preview: 57.3 @ 66Gemini 2.5 Flash Lite: 57 @ 6Qwen3 4B: 56.5 @ 103Sarvam M: 56.4 @ 136GPT-4o: 56.4 @ 132Reka Flash 3: 56.2 @ 93Llama 3.1 70B Instruct: 56 @ 1204Llama 3.1 70B InstructCommand A: 55.9 @ 203Gemma 3 12B: 55.5 @ 33Qwen3 Coder 30B A3B Instruct: 55.4 @ 97Claude 3.5 Haiku: 54.5 @ 104Gemini 2.0 Flash Lite: 54.4 @ 85Devstral 2: 54.3 @ 51Llama 3.2 90B Instruct: 54 @ 100Nova Premier: 53.1 @ 40Pixtral Large: 53.1 @ 0Reka Flash: 52.9 @ 85Mistral Medium 3.1: 51.5 @ 47Mistral Small 3.2: 51 @ 100Mistral Small 3.1 24B Base: 50.9 @ 137Llama 3.3 70B Instruct: 50.6 @ 2220Llama 3.3 70B InstructQwen2.5 Turbo: 50.3 @ 67Qwen2.5 Coder 32B Instruct: 50.1 @ 110Qwen3 VL 8B: 49.7 @ 120GPT-4o-mini: 49.1 @ 92Nova Micro: 49 @ 100Gemini 1.5 Flash 8B: 48.4 @ 150Ministral 3 14B: 47.9 @ 67Mistral Large 2: 47.9 @ 42Devstral Medium: 47.8 @ 72Phi 4: 47.6 @ 33Devstral Small 2: 47.5 @ 62LFM2-24B-A2B: 47.4 @ 208Qwen3 1.7B: 46.9 @ 138Nemotron 3 Nano Omni 30B A3B Reasoning: 46.9 @ 301Qwen2.5 7B Instruct: 46.6 @ 138Phi-4-multimodal-instruct: 46.2 @ 25Phi-3.5-mini-instruct: 46 @ 23Claude 3 Sonnet: 45.8 @ 120Gemma 3 4B: 45.8 @ 33Qwen3.5 2B: 45.6 @ 328Devstral Small: 45.3 @ 190Llama 3.1 8B Instruct: 45 @ 2047Llama 3.1 8B InstructLlama 3.1 Nemotron 70B Instruct: 43.7 @ 292Mistral Small 3: 43.6 @ 136Ministral 3 8B: 43.3 @ 86Granite 4.1 8B: 43.3 @ 133Qwen3 VL 8B Instruct: 43 @ 145Jamba 1.5 Large: 42.8 @ 100Jamba 1.6 Large: 42.6 @ 52Hermes 3 - Llama-3.1 70B: 42.5 @ 32Mistral Small 3.1: 42.4 @ 134GPT-4.1 Nano: 42.1 @ 200Llama 3 70B Instruct: 40.8 @ 45Mistral Small: 40.3 @ 134Claude 3 Haiku: 38.9 @ 104Llama 3.2 11B Instruct: 36.7 @ 168Jamba Large 1.7: 36.5 @ 48Granite 4.0 H Small: 35.7 @ 524GPT-3.5 Turbo: 35.2 @ 100Gemma 3n E4B Instruct: 34.7 @ 56Gemini 1.0 Pro: 34 @ 120Ministral 3 3B: 33.7 @ 154Mistral Medium: 33.6 @ 45Solar Mini: 33.1 @ 63Granite 3.3 8B: 32.5 @ 376Llama 3 8B Instruct: 32.4 @ 81Llama 3.2 3B Instruct: 30.8 @ 172Tiny Aya Global: 30.5 @ 126Jamba 1.5 Mini: 29.6 @ 100Qwen3 0.6B: 29.3 @ 225Command R+: 28.9 @ 100Jamba 1.6 Mini: 24.9 @ 183Gemma 3n E4B Instructed: 24.8 @ 42Qwen3.5 0.8B: 23.6 @ 120Mistral 7B Instruct: 14.7 @ 90Llama 3.2 1B Instruct: 12.1 @ 91Llama 2 Chat 7B: 11.3 @ 113
#ModelMulti idxAI2DMMMU-ProChartQADocVQAMathVistaMMMUContextSpeedIn $/M
1Claude 3.5 Sonnet83.394.790.895.267.768.3200K101$3.00
2Gemma 3 27B8384.57886.6131K33$0.08
3o4-mini82.984.381.6200K115$1.10
4Gemma 3 12B82.384.275.787.1131K33$0.04
5Gemini 2.5 Pro Preview 06-0582821M85$1.25
6o38276.486.882.9200K50$2.00
7Pixtral Large81.793.888.193.369.464131K0$2.00
8Nova Pro81.589.293.561.7300K100$0.80
9GPT-581.378.484.2400K100$1.25
10Mistral Small 3.2 24B Instruct8192.987.494.967.162.5
11Llama 4 Scout80.888.894.470.769.410M776$0.08
12Gemini 2.5 Flash79.779.71M85$0.30
13Gemini 2.5 Pro79.679.61M85$1.25
14Qwen2.5 VL 72B Instruct79.188.451.189.596.470.2131K$0.25
15Nova Lite78.586.892.456.2300K100$0.06
16Llama 4 Maverick78.259.69094.473.773.41M639$0.15
17Grok-37878128K100$3.00
18GPT-4o77.794.259.985.792.861.472.2128K132$2.50
19Claude Opus 4.677.377.31M48$5.00
20Grok-276.293.66966.1128K85$2.00
21Gemini 2.0 Flash Thinking75.475.4$0.00
22Claude 3.7 Sonnet7575200K101$3.00
23DeepSeek VL274.981.48693.362.851.1129K22$9.50
24Grok-2 mini74.893.268.163.2
25o174.771.877.6200K66$15.00
26Claude Sonnet 474.474.41M101$3.00
27GPT-4.573.872.375.2128K50$75.00
28GPT-4.173.572.274.81M100$2.00
29DeepSeek VL2 Small73.18084.592.360.748
30Gemma 3 4B73.174.868.875.8131K33$0.04
31GPT-4.1 Mini72.973.172.71M150$0.40
32Gemini 2.5 Flash Lite72.972.91M6$0.10
33Kimi-k1.572.574.970
34Llama 3.2 90B Instruct71.892.345.285.590.157.360.3128K100$0.35
35Qwen2.5 VL 32B Instruct71.449.594.870
36Grok-1.5V71.388.376.185.652.853.6
37Qwen2.5-Omni-7B71.283.236.685.395.267.959.2
38QvQ-72B-Preview70.971.470.3
39Pixtral-12B70.881.890.75852.5128K0$0.15
40Gemini 2.0 Flash70.770.71M183$0.10
41Qwen2.5 VL 7B Instruct7038.387.395.758.6
42Phi-4-multimodal-instruct68.882.338.581.493.262.455.1128K25$0.05
43Gemini 2.0 Flash Lite68681M85$0.08
44Qwen2-VL-72B-Instruct67.346.288.3
45DeepSeek VL2 Tiny67.271.68188.953.640.7
46Gemini 1.5 Pro6768.165.92M85$1.25
47Llama 3.2 11B Instruct66.491.13383.488.451.550.7128K168$0.05
48Gemini 1.5 Flash64.165.862.31M150$0.15
49Grok-1.56485.652.853.6
50Phi-3.5-vision-instruct61.778.181.843.943
51Mistral Small 3.1 24B Base59.359.3128K137$0.10
52Mistral Small 3.1 24B Instruct59.359.3
53GPT-4o-mini58.156.759.4128K92$0.15
54GPT-4.1 Nano55.856.255.41M200$0.10
55Gemini 1.5 Flash 8B54.254.753.71M150$0.07
56Gemini 1.0 Pro47.346.647.933K120$0.50
57GPT-3.5 Turbo00016K100$0.50

57 models ranked on Multimodal. The intelligence index is a balanced mean of per-category scores; category columns average the benchmarks within each. Scores are curated approximations — see each model for sources. Click any column to sort.