LLM rankings, places 76 to 150
- models measured
- 123
- lead over #2
- +2.4
- median
- 53.5
Leader on LLMBenchmark Index, a weighted mean of 7 inputs, 90% margin · read October 21
As of October 21, 2026, Anthropic Claude Opus 5.5 leads the LLMBenchmark Index of 123 ranked models with 87.1 ±2.1. 56 more models have only two of the seven inputs and are listed below them without a rank. The Artificial Analysis Intelligence Index is a column of its own; every other column is one category read from its own public board, and a model a board never measured says "No data". Nothing is estimated.
75 of 179 shown
JSON| Rank | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| 76 | R1 0528 DeepSeek | 46.8 ±17.9 | 13.1 | No data | No data | 37.4 | 66.5 | No data | 1149 | $2.18 |
| 77 | T Inkling Thinking Machines | 46.5 ±12.6 | 25.0 | 31.5 | 14.6 | 63.1 | 64.8 | 56.7 | 1224 | $4.05 |
| 78 | GPT-5.4 Nano OpenAI | 45.7 ±9.2 | 20.7 | 44.8 | No data | 46.0 | 66.6 | 26.3 | No data | $1.25 |
| 79 | Qwen3.6 27B Qwen | 45.3 ±11.1 | 21.4 | 36.8 | No data | 59.4 | 63.1 | No data | No data | $2.70 |
| 80 | Qwen3.6 35B A3B Qwen | 45.0 ±14.1 | 18.2 | No data | No data | 58.3 | 53.5 | No data | No data | $1.00 |
| 81 | Muse Spark Open Meta | 44.9 ±17.7 | No data | 47.4 | 19.4 | 67.1 | No data | No data | 1317 | $4.25 |
| 82 | Kimi K2.5 Moonshot AI | 44.8 ±11.7 | 23.5 | 38.6 | No data | 31.0 | No data | 50.8 | 1248 | $2.85 |
| 83 | o1 OpenAI | 44.6 ±10.3 | 15.2 | No data | No data | 43.4 | 48.2 | 52.8 | No data | $60.00 |
| 84 | MiniMax M2.7 MiniMax | 44.5 ±14.3 | 22.8 | 36.2 | No data | No data | No data | No data | 1243 | $1.20 |
| 85 | GPT-5.4 Mini OpenAI | 44.0 ±11.5 | 24.1 | 35.8 | 13.0 | 52.4 | 70.0 | 46.1 | No data | $4.50 |
| 86 | GPT-5 Mini OpenAI | 43.7 ±9.0 | 20.6 | 34.2 | No data | 40.8 | 65.9 | 38.3 | 1125 | $2.00 |
| 87 | T Inkling Small Thinking Machines | 43.6 ±12.4 | 25.7 | 27.3 | 14.5 | 61.1 | 69.8 | 40.2 | No data | $1.20 |
| 88 | o4 Mini OpenAI | 43.0 ±10.4 | 16.7 | No data | No data | 40.2 | 60.1 | 35.2 | 985 | $4.40 |
| 89 | Gemma 4 31B Google | 42.1 ±14.7 | 14.7 | 39.0 | 8.7 | 58.1 | 74.6 | 26.0 | No data | $0.40 |
| 90 | GLM 4.7 Flash Z.ai | 42.1 ±13.0 | 14.9 | No data | No data | 35.7 | 58.6 | No data | 1194 | $0.40 |
| 91 | Gemini 2.5 Pro Google | 41.8 ±16.2 | 16.1 | 19.1 | No data | 45.3 | 58.7 | No data | 1167 | $10.00 |
| 92 | 41.5 ±18.3 | 7.5 | No data | No data | 35.2 | 63.4 | No data | No data | $0.30 | |
| 93 | Claude Haiku 4.5 Anthropic | 41.0 ±10.7 | 16.9 | 35.6 | 19.7 | 49.3 | 66.5 | 32.9 | 1122 | $5.00 |
| 94 | Grok 4.20 xAI | 41.0 ±9.2 | 25.7 | 33.3 | No data | 55.6 | No data | No data | 1229 | $2.50 |
| 95 | Qwen3 32B Qwen | 40.8 ±17.4 | 7.3 | No data | No data | 38.1 | 66.8 | No data | No data | $0.40 |
| 96 | MiMo-V2.5-Pro Xiaomi | 39.3 ±18.5 | 26.0 | 31.6 | 13.5 | No data | No data | No data | 1267 | $0.9152 |
| 97 | gpt-oss-20b OpenAI | 38.5 ±13.7 | 10.0 | No data | No data | 38.3 | 61.1 | No data | 852 | $0.15 |
| 98 | Qwen3 30B A3B Qwen | 38.2 ±15.7 | 7.6 | No data | No data | 35.3 | 61.6 | No data | 955 | $0.50 |
| 99 | o3 Mini OpenAI | 38.0 ±10.6 | 12.5 | No data | No data | 36.1 | 51.4 | 30.3 | No data | $4.40 |
| 100 | DeepSeek V3.2 DeepSeek | 37.9 ±15.0 | 21.5 | 30.0 | No data | 22.7 | No data | No data | 1175 | $0.88 |
| 101 | MiniMax M2.5 MiniMax | 37.3 ±11.5 | 22.8 | 37.2 | No data | 19.5 | No data | No data | 1222 | $1.20 |
| 102 | Mistral Medium 3.5 Mistral AI | 37.3 ±20.5 | 14.2 | 25.5 | No data | No data | No data | No data | 1142 | $7.50 |
| 103 | 36.6 ±20.4 | 15.6 | 22.7 | No data | No data | No data | No data | 1083 | $1.50 | |
| 104 | GPT-4.1 OpenAI | 36.5 ±12.3 | 12.7 | No data | No data | 27.4 | 30.0 | 42.9 | 1039 | $8.00 |
| 105 | Nemotron 3.5 Super VL NVIDIA | 35.6 ±18.2 | 26.5 | No data | 11.1 | 57.8 | No data | No data | 1172 | $0.50 |
| 106 | GPT-5 Nano OpenAI | 35.6 ±10.3 | 13.0 | No data | No data | 31.8 | 51.0 | 25.1 | 1101 | $0.40 |
| 107 | Gemini 2.5 Flash Google | 34.3 ±24.0 | 15.5 | No data | No data | 11.4 | No data | No data | 1074 | $2.50 |
| 108 | GLM 4.6 Z.ai | 33.4 ±25.7 | 18.5 | 8.8 | No data | No data | No data | No data | 1174 | $2.20 |
| 109 | GPT-4.1 Mini OpenAI | 31.9 ±10.9 | 10.2 | No data | No data | 27.4 | 32.0 | 27.0 | 997 | $1.60 |
| 110 | Gemma 3 27B Google | 31.4 ±16.1 | 4.9 | No data | No data | 27.0 | 32.3 | No data | No data | $0.30 |
| 111 | Gemma 3 12B Google | 28.0 ±15.5 | 3.8 | No data | No data | 22.7 | 28.2 | No data | No data | $0.1625 |
| 112 | GPT-4o (2024-05-13) OpenAI | 27.1 ±12.5 | 7.3 | No data | No data | 25.7 | 20.0 | No data | No data | $15.00 |
| 113 | Qwen2.5 72B Instruct Qwen | 26.5 ±10.5 | 7.7 | No data | No data | 27.1 | 21.1 | No data | No data | $0.40 |
| 114 | GPT-4o (2024-08-06) OpenAI | 26.3 ±10.2 | 7.7 | No data | No data | 26.2 | 12.2 | 32.6 | No data | $10.00 |
| 115 | Llama 3.3 70B Instruct Meta Llama | 26.0 ±11.0 | 7.7 | No data | No data | 26.3 | 18.3 | No data | No data | $0.72 |
| 116 | Llama 4 Scout Meta Llama | 24.9 ±11.6 | 6.5 | No data | No data | 21.5 | 21.1 | No data | 750 | $0.50 |
| 117 | Phi 4 Microsoft | 24.7 ±8.9 | 5.9 | No data | No data | 30.7 | 22.8 | No data | No data | $0.14 |
| 118 | GPT-4.1 Nano OpenAI | 24.7 ±11.0 | 7.8 | No data | No data | 20.4 | 34.5 | 7.8 | 973 | $0.40 |
| 119 | Llama 3.1 70B Instruct Meta Llama | 24.3 ±10.8 | 6.6 | No data | No data | 25.3 | 16.4 | No data | No data | $0.56 |
| 120 | Gemma 3 4B Google | 22.6 ±13.5 | 4.8 | No data | No data | 15.4 | 18.3 | No data | No data | $0.075 |
| 121 | Llama 4 Maverick Meta Llama | 21.8 ±17.9 | 9.3 | No data | No data | 4.2 | No data | No data | 871 | $1.00 |
| 122 | Llama 3.1 8B Instruct Meta Llama | 17.9 ±7.1 | 6.9 | No data | No data | 17.3 | 11.1 | No data | No data | $0.08 |
| 123 | Llama 3.2 1B Instruct Meta Llama | 9.6 ±3.4 | 4.8 | No data | No data | 15.9 | 5.1 | No data | No data | $0.201 |
| thin | Muse Wonder Meta | 76.9 ±6.7 Thin evidence: 2 of 7 inputs | No data | 71.9 | No data | No data | No data | No data | 1362 | $20.00 |
| thin | GPT-5.3-Codex OpenAI | 54.2 ±7.8 Thin evidence: 2 of 7 inputs | 32.5 | 60.0 | No data | No data | No data | No data | 1163 | $14.00 |
| thin | Gemini 3.1 Flash Lite Google | 53.5 ±0.4 Thin evidence: 2 of 7 inputs | No data | No data | No data | 53.2 | 53.9 | No data | No data | $1.50 |
| thin | DeepSeek V4 Flash 0423 DeepSeek | 53.4 ±21.6 Thin evidence: 2 of 7 inputs | 24.4 | No data | No data | No data | No data | No data | 1208 | $0.28 |
| thin | GLM 5V Turbo Z.ai | 53.2 ±22.9 Thin evidence: 2 of 7 inputs | 23.5 | No data | No data | No data | No data | No data | 1230 | $4.00 |
| thin | Solar Pro 4 Upstage | 52.5 ±14.2 Thin evidence: 2 of 7 inputs | 28.2 | No data | No data | No data | No data | No data | 1176 | $0.60 |
| thin | Hy3 preview Tencent | 51.5 ±17.9 Thin evidence: 2 of 7 inputs | 25.3 | No data | No data | No data | No data | No data | No data | $0.60 |
| thin | Qwen3.5-27B Qwen | 50.4 ±20.6 Thin evidence: 2 of 7 inputs | 22.9 | No data | No data | No data | No data | No data | No data | $1.56 |
| thin | Qwen3.5-122B-A10B Qwen | 47.4 ±25.8 Thin evidence: 2 of 7 inputs | 17.7 | No data | No data | No data | No data | No data | No data | $2.08 |
| thin | DeepSeek V3.2 Exp DeepSeek | 47.3 ±27.4 Thin evidence: 2 of 7 inputs | 16.6 | No data | No data | No data | No data | No data | 1179 | $0.41 |
| thin | Step 3.5 Flash StepFun | 45.7 ±24.9 Thin evidence: 2 of 7 inputs | 17.0 | No data | No data | No data | No data | No data | No data | $0.30 |
| thin | DeepSeek V3.1 Terminus DeepSeek | 45.5 ±28.3 Thin evidence: 2 of 7 inputs | 14.8 | No data | No data | No data | No data | No data | 1187 | $1.0142 |
| thin | GLM 4.5 Z.ai | 45.0 ±31.0 Thin evidence: 2 of 7 inputs | 12.8 | No data | No data | No data | No data | No data | 1170 | $2.20 |
| thin | Qwen-Plus Qwen | 44.6 ±13.9 Thin evidence: 2 of 7 inputs | No data | No data | No data | No data | 30.0 | No data | No data | $0.78 |
| thin | GPT-5.2-Codex OpenAI | 43.4 ±2.7 Thin evidence: 2 of 7 inputs | 28.5 | 45.4 | No data | No data | No data | No data | No data | $14.00 |
| thin | 42.1 ±32.5 Thin evidence: 2 of 7 inputs | 9.9 | No data | No data | No data | No data | No data | No data | $1.90 | |
| thin | Mistral Medium 3.1 Mistral AI | 41.9 ±33.5 Thin evidence: 2 of 7 inputs | 9.2 | No data | No data | No data | No data | No data | 1133 | $2.00 |
| thin | 41.7 ±3.7 Thin evidence: 2 of 7 inputs | No data | 39.5 | No data | No data | No data | 46.2 | No data | $1.80 | |
| thin | MiniMax M2 MiniMax | 41.6 ±17.5 Thin evidence: 2 of 7 inputs | 18.6 | No data | No data | No data | No data | No data | 1143 | $1.02 |
| thin | 41.6 ±32.4 Thin evidence: 2 of 7 inputs | 9.6 | No data | No data | No data | No data | No data | No data | $1.10 | |
| thin | Nemotron 3 Super NVIDIA | 40.6 ±26.0 Thin evidence: 2 of 7 inputs | 12.8 | No data | No data | No data | No data | No data | No data | $0.45 |
| thin | GLM 4.5 Air Z.ai | 39.9 ±28.0 Thin evidence: 2 of 7 inputs | 11.1 | No data | No data | No data | No data | No data | 1146 | $1.10 |
| thin | Nova 2 Lite Amazon | 39.7 ±23.9 Thin evidence: 2 of 7 inputs | 13.4 | No data | No data | No data | No data | No data | No data | $2.50 |
| thin | GLM 4.6V Z.ai | 39.4 ±27.2 Thin evidence: 2 of 7 inputs | 11.2 | No data | No data | No data | No data | No data | No data | $0.90 |
| thin | 38.7 ±26.4 Thin evidence: 2 of 7 inputs | 11.2 | No data | No data | No data | No data | No data | No data | $1.20 | |
| thin | 38.2 ±24.7 Thin evidence: 2 of 7 inputs | 11.9 | No data | No data | No data | No data | No data | 1159 | $1.50 | |
| thin | Mercury 2 Inception | 37.3 ±25.4 Thin evidence: 2 of 7 inputs | 10.8 | No data | No data | No data | No data | No data | 1036 | $0.75 |
First on that board Top 5% Top 15% Top 35% of the models it measured
LLMBenchmark Index: our overall score, 0 to 100, with a 90% margin, a weighted mean of our five indexes, the Artificial Analysis Intelligence Index and the LMArena text Elo; three of the seven are needed to be ranked, how it is built. AA Intelligence Index: Artificial Analysis, as published. Coding: our Vibe Coding Index, with its benchmarks as dots. Agentic, Reasoning, Math and Knowledge: our indexes, each a weighted mean of three or four live benchmarks, two needed to be scored. Web design: Design Arena, websites, Elo. Price: OpenRouter, per million output tokens.