GPQA Diamond (Artificial Analysis run) leaderboard
- GPT-6 Astra leads
- 96.3%
- models measured
- 175
- lead over #2
- +1.0 pts
- median
- 81.7%
Leader on GPQA Diamond (Artificial Analysis run) · read October 21
On October 21, 2026, 175 of our catalog's models had a GPQA Diamond (Artificial Analysis run) result, and OpenAI GPT-6 Astra led with 96.3%.
Graduate-level science questions written to resist web search. Share answered correctly. Run by Artificial Analysis on its own harness.
Role in the index
We collect this benchmark and show it as supporting evidence. It is not part of the Vibe Coding Index, because it measures something other than coding and the index is about coding. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-6 AstraOpenAI | 96.3% | 88.7 | $50.00 | |
| 2 | Gemini 3.8 FlashGoogle | 95.3% | 70.5 | $4.125 | |
| 3 | Grok 4.6xAI | 94.9% | 76.4 | $6.00 | |
| 4 | Gemini 3.7 FlashGoogle | 94.5% | 69.9 | $3.75 | |
| 5 | Gemini 3.1 Pro PreviewGoogle | 94.1% | 35.3 | $12.00 | |
| 6 | Muse Spark 1.3Meta | 94.1% | 79.7 | $4.25 | |
| 7 | GPT-5.6 SolOpenAI | 94.1% | 76.7 | $20.00 | |
| 8 | Claude Fable 5.1Anthropic | 93.7% | 89.1 | $50.00 | |
| 9 | Claude Opus 5Anthropic | 93.7% | 81.8 | $25.00 | |
| 10 | Kimi K3Moonshot AI | 93.5% | 61.8 | $11.936 | |
| 11 | GPT-5.5OpenAI | 93.5% | 67.8 | $30.00 | |
| 12 | Qwen3.8 2.4T A95BQwen | 93.5% | Not benchmarked | $6.00 | |
| 13 | Grok 4.5xAI | 93.1% | 75.5 | $6.00 | |
| 14 | MiniMax M3MiniMax | 92.9% | 57.8 | $1.20 | |
| 15 | DeepSeek V4 Pro 0813DeepSeek | 92.8% | 54.6 | $1.98 | |
| 16 | Gemini 3.6 FlashGoogle | 92.8% | 48.3 | $3.75 | |
| 17 | Qwen3.8 Max (0902)Qwen | 92.8% | 63.5 | $6.00 | |
| 18 | Claude Fable 5Anthropic | 92.6% | 82.1 | $50.00 | |
| 19 | GPT-5.6 TerraOpenAI | 92.5% | 69.0 | $12.00 | |
| 20 | Qwen3.7 MaxQwen | 92.3% | 45.1 | $4.425 | |
| 21 | Gemini 3.5 FlashGoogle | 92.2% | 42.8 | $9.00 | |
| 22 | Claude Opus 4.8Anthropic | 92.0% | 66.0 | $25.00 | |
| 23 | GPT-5.4OpenAI | 92.0% | 37.7 | $15.00 | |
| 24 | GLM 5.3Z.ai | 91.7% | 73.1 | $1.40 | |
| 25 | GPT-5.3-CodexOpenAI | 91.5% | 29.5 | $14.00 | |
| 26 | Claude Opus 4.7Anthropic | 91.4% | 51.1 | $25.00 | |
| 27 | DeepSeek V4 Flash Vision ExpDeepSeek | 91.3% | 37.6 | $0.6468 | |
| 28 | GLM 5.3 FlashZ.ai | 91.2% | 71.0 | $0.50 | |
| 29 | Claude Sonnet 5Anthropic | 91.1% | 63.5 | $10.00 | |
| 30 | Kimi K2.6Moonshot AI | 91.1% | 44.2 | $3.41 | |
| 31 | GPT-5.6 LunaOpenAI | 91.1% | 61.5 | $1.20 | |
| 32 | Grok 4.20xAI | 91.1% | Not benchmarked | $2.50 | |
| 33 | DeepSeek V4 Flash 0731DeepSeek | 90.8% | Not benchmarked | $1.32 | |
| 34 | Qwen3.8 27BQwen | 90.5% | 73.6 | $1.35 | |
| 35 | Muse Spark 1.2Meta | 90.4% | 65.1 | $4.25 | |
| 36 | GPT-5.2OpenAI | 90.3% | 58.9 | $14.00 | |
| 37 | Grok 4.3xAI | 90.1% | 22.5 | $2.50 | |
| 38 | Qwen3.7 PlusQwen | 90.0% | Not benchmarked | $1.28 | |
| 39 | GPT-5.2-CodexOpenAI | 89.9% | 55.1 | $14.00 | |
| 40 | Gemini 3 Flash PreviewGoogle | 89.8% | 75.8 | $3.00 | |
| 41 | Muse Spark 1.1Meta | 89.8% | 48.9 | $4.25 | |
| 42 | Hy3Tencent | 89.7% | 43.9 | $0.528 | |
| 43 | Claude Opus 4.6Anthropic | 89.6% | 66.9 | $25.00 | |
| 44 | Kimi K2.7 CodeMoonshot AI | 89.6% | 39.1 | $3.35 | |
| 45 | T Inkling SmallThinking Machines | 89.5% | 29.9 | $1.20 | |
| 46 | GLM 5.2Z.ai | 89.5% | 57.7 | $7.00 | |
| 47 | Qwen3.5 397B A17BQwen | 89.3% | 28.6 | $2.34 | |
| 48 | Solar Pro 4Upstage | 89.1% | 24.4 | $0.60 | |
| 49 | Muse Spark OpenMeta | 89.0% | 62.1 | $4.25 | |
| 50 | Qwen3.6 PlusQwen | 88.2% | 37.4 | $1.95 | |
| 51 | Kimi K2.5Moonshot AI | 87.9% | 58.5 | $2.50 | |
| 52 | Claude Sonnet 4.6Anthropic | 87.5% | 57.8 | $15.00 | |
| 53 | GPT-5.4 MiniOpenAI | 87.5% | 28.2 | $4.50 | |
| 54 | MiniMax M2.7MiniMax | 87.4% | 28.4 | $1.20 | |
| 55 | GPT-5.1OpenAI | 87.3% | 53.3 | $10.00 | |
| 56 | T InklingThinking Machines | 87.2% | 49.7 | $4.05 | |
| 57 | GLM 5.1Z.ai | 86.8% | 44.0 | $4.40 | |
| 58 | Nemotron 3 UltraNVIDIA | 86.7% | Not benchmarked | $2.40 | |
| 59 | Hy3 previewTencent | 86.7% | 22.4 | $0.60 | |
| 60 | Claude Opus 4.5Anthropic | 86.6% | 66.8 | $25.00 | |
| 61 | MiMo-V2.5-ProXiaomi | 86.6% | 40.0 | $0.609 | |
| 62 | Ling 3.0 Flash VLinclusionAI | 86.2% | Not benchmarked | $0.18 | |
| 63 | GPT-5.1-CodexOpenAI | 86.0% | 50.5 | $10.00 | |
| 64 | GLM 4.7Z.ai | 85.9% | 33.6 | $1.98 | |
| 65 | Qwen3.5-27BQwen | 85.8% | 22.6 | $2.16 | |
| 66 | Gemma 4 31BGoogle | 85.7% | 38.3 | $0.40 | |
| 67 | Qwen3.5-122B-A10BQwen | 85.7% | 22.9 | $2.08 | |
| 68 | Ling 3.0 FlashinclusionAI | 85.5% | Not benchmarked | $0.063 | |
| 69 | GPT-5OpenAI | 85.4% | 72.2 | $10.00 | |
| 70 | MiMo-V2.5Xiaomi | 84.9% | 34.0 | $0.238 | |
| 71 | MiniMax M2.5MiniMax | 84.8% | 59.4 | $0.95 | |
| 72 | GLM 5 TurboZ.ai | 84.7% | Not benchmarked | $4.00 | |
| 73 | o3 ProOpenAI | 84.5% | 84.9 | $80.00 | |
| 74 | Qwen3.5-35B-A3BQwen | 84.5% | 7.7 | $1.00 | |
| 75 | Gemini 2.5 ProGoogle | 84.4% | 37.0 | $10.00 | |
| 76 | Qwen3.6 27BQwen | 84.2% | Not benchmarked | $2.00 | |
| 77 | Qwen3.6 35B A3BQwen | 84.1% | Not benchmarked | $0.95 | |
| 78 | DeepSeek V3.2DeepSeek | 84.0% | 54.4 | $0.42 | |
| 79 | Gemini 3.5 Flash LiteGoogle | 83.8% | 33.6 | $2.50 | |
| 80 | Kimi K2 ThinkingMoonshot AI | 83.8% | 63.4 | $2.50 | |
| 81 | Gemini 2.5 Pro Preview 06-05Google | 83.6% | 83.1 | $10.00 | |
| 82 | Step 3.5 FlashStepFun | 83.1% | Not benchmarked | $0.30 | |
| 83 | MiniMax M2.1MiniMax | 83.0% | Not benchmarked | $1.20 | |
| 84 | GPT-5 MiniOpenAI | 82.8% | 59.8 | $2.00 | |
| 85 | o3OpenAI | 82.7% | 70.7 | $8.00 | |
| 86 | Gemini 3.1 Flash LiteGoogle | 82.2% | Not benchmarked | $1.50 | |
| 87 | GLM 5Z.ai | 82.0% | 59.7 | $3.20 | |
| 88 | GPT-5.4 NanoOpenAI | 81.7% | Not benchmarked | $1.25 | |
| 89 | R1 0528DeepSeek | 81.3% | 71.4 | $2.15 | |
| 90 | GPT-5.1-Codex-MiniOpenAI | 81.3% | 6.2 | $2.00 | |
| 91 | Step 3.7 FlashStepFun | 80.9% | Not benchmarked | $1.15 | |
| 92 | GLM 5V TurboZ.ai | 80.9% | 28.8 | $4.00 | |
| 93 | Qwen3.5-9BQwen | 80.6% | Not benchmarked | $0.13 | |
| 94 | Nemotron 3 SuperNVIDIA | 80.0% | Not benchmarked | $2.40 | |
| 95 | DeepSeek V3.2 ExpDeepSeek | 79.7% | 55.0 | $0.41 | |
| 96 | DeepSeek V3.1 TerminusDeepSeek | 79.2% | Not benchmarked | $1.00 | |
| 97 | Gemma 4 26B A4BGoogle | 79.2% | 22.8 | $0.255 | |
| 98 | Gemini 2.5 FlashGoogle | 79.0% | 28.7 | $2.50 | |
| 99 | Qwen3 235B A22B Instruct 2507Qwen | 79.0% | Not benchmarked | $0.55 | |
| 100 | o4 MiniOpenAI | 78.4% | 68.6 | $4.40 | |
| 101 | gpt-oss-120bOpenAI | 78.2% | 34.5 | $0.60 | |
| 102 | GLM 4.5Z.ai | 78.2% | 64.2 | $2.20 | |
| 103 | LongCat 2.0Meituan | 78.0% | Not benchmarked | $1.20 | |
| 104 | GLM 4.6Z.ai | 78.0% | 52.2 | $2.20 | |
| 105 | MiniMax M2MiniMax | 77.7% | 45.3 | $1.02 | |
| 106 | o3 MiniOpenAI | 77.3% | 52.1 | $4.40 | |
| 107 | Qwen3 VL 235B A22B InstructQwen | 77.2% | Not benchmarked | $1.90 | |
| 108 | Mercury 2Inception | 77.0% | 0.0 | $0.75 | |
| 109 | Mistral Small 4Mistral AI | 76.9% | Not benchmarked | $0.60 | |
| 110 | Kimi K2 0905Moonshot AI | 76.7% | Not benchmarked | $2.50 | |
| 111 | Kimi K2 0711Moonshot AI | 76.6% | 64.7 | $2.30 | |
| 112 | Command ACohere | 76.1% | Not benchmarked | $10.00 | |
| 113 | Qwen3 Next 80B A3B InstructQwen | 75.9% | Not benchmarked | $1.10 | |
| 114 | Nemotron 3 Nano 30B A3BNVIDIA | 75.7% | Not benchmarked | $0.24 | |
| 115 | Trinity Large ThinkingArcee AI | 75.2% | 5.4 | $0.80 | |
| 116 | Mistral Medium 3.5Mistral AI | 74.8% | 9.2 | $7.50 | |
| 117 | o1OpenAI | 74.7% | 61.7 | $60.00 | |
| 118 | Nemotron 3.5 LightningNVIDIA | 74.3% | Not benchmarked | $0.20 | |
| 119 | Qwen3 Coder NextQwen | 73.7% | Not benchmarked | $0.80 | |
| 120 | GLM 4.5 AirZ.ai | 73.3% | Not benchmarked | $0.85 | |
| 121 | Solar Pro 3Upstage | 72.4% | Not benchmarked | $0.60 | |
| 122 | Qwen3 VL 30B A3B InstructQwen | 72.0% | Not benchmarked | $0.60 | |
| 123 | GLM 4.6VZ.ai | 71.9% | Not benchmarked | $0.90 | |
| 124 | Qwen3 30B A3B Instruct 2507Qwen | 70.7% | Not benchmarked | $0.30 | |
| 125 | gpt-oss-20bOpenAI | 68.8% | Not benchmarked | $0.09 | |
| 126 | GLM 4.5VZ.ai | 68.4% | Not benchmarked | $1.80 | |
| 127 | GPT-5 NanoOpenAI | 67.6% | 34.8 | $0.40 | |
| 128 | Llama 4 MaverickMeta Llama | 67.1% | 18.1 | $0.6525 | |
| 129 | Qwen3 32BQwen | 66.8% | 40.0 | $0.28 | |
| 130 | GPT-4.1OpenAI | 66.6% | 46.5 | $8.00 | |
| 131 | GPT-4.1 MiniOpenAI | 66.4% | 28.5 | $1.60 | |
| 132 | DeepSeek V3 0324DeepSeek | 65.5% | 42.0 | $1.14 | |
| 133 | Granite 4.2 8BIBM Granite | 63.1% | Not benchmarked | $0.25 | |
| 134 | Gemini 2.5 Flash LiteGoogle | 62.5% | Not benchmarked | $0.40 | |
| 135 | Qwen3 Coder 480B A35BQwen | 61.8% | 50.0 | $1.00 | |
| 136 | Qwen3 30B A3BQwen | 61.6% | Not benchmarked | $0.50 | |
| 137 | Qwen3 14BQwen | 60.4% | Not benchmarked | $0.24 | |
| 138 | Mistral Medium 3.1Mistral AI | 58.8% | Not benchmarked | $2.00 | |
| 139 | Llama 4 ScoutMeta Llama | 58.7% | 9.1 | $0.30 | |
| 140 | GLM 4.7 FlashZ.ai | 58.1% | Not benchmarked | $0.40 | |
| 141 | Qwen3 VL 8B InstructQwen | 57.9% | Not benchmarked | $0.75 | |
| 142 | Mistral Medium 3Mistral AI | 57.8% | Not benchmarked | $2.00 | |
| 143 | Sonar ProPerplexity | 57.8% | Not benchmarked | $15.00 | |
| 144 | Phi 4Microsoft | 57.5% | Not benchmarked | $0.14 | |
| 145 | Nova Premier 1.0Amazon | 56.9% | Not benchmarked | $12.50 | |
| 146 | GPT-4oOpenAI | 54.3% | 38.8 | $10.00 | |
| 147 | Reka Flash 3Reka | 52.9% | Not benchmarked | $0.20 | |
| 148 | GPT-4o (2024-05-13)OpenAI | 52.6% | Not benchmarked | $15.00 | |
| 149 | GPT-4o (2024-08-06)OpenAI | 52.1% | 23.1 | $10.00 | |
| 150 | Qwen3 Coder 30B A3B InstructQwen | 51.6% | 60.4 | $0.28 | |
| 151 | GPT-4.1 NanoOpenAI | 51.2% | 8.9 | $0.40 | |
| 152 | Nova Pro 1.0Amazon | 49.9% | Not benchmarked | $3.20 | |
| 153 | Llama 3.3 70B InstructMeta Llama | 49.8% | Not benchmarked | $0.32 | |
| 154 | Qwen2.5 72B InstructQwen | 49.1% | Not benchmarked | $0.40 | |
| 155 | Mistral Large 2407Mistral AI | 47.2% | Not benchmarked | $6.00 | |
| 156 | SonarPerplexity | 47.1% | Not benchmarked | $1.00 | |
| 157 | Mistral Small 3Mistral AI | 46.2% | Not benchmarked | $0.08 | |
| 158 | Nova Lite 1.0Amazon | 43.3% | Not benchmarked | $0.24 | |
| 159 | Gemma 3 27BGoogle | 42.8% | 4.9 | $0.35 | |
| 160 | GPT-4o-miniOpenAI | 42.6% | Not benchmarked | $0.60 | |
| 161 | SabaMistral AI | 42.4% | Not benchmarked | $0.60 | |
| 162 | Qwen2.5 Coder 32B InstructQwen | 41.7% | 30.5 | $1.00 | |
| 163 | Llama 3.1 70B InstructMeta Llama | 40.9% | Not benchmarked | $0.40 | |
| 164 | N Hermes 3 70B InstructNous Research | 40.1% | Not benchmarked | $0.70 | |
| 165 | Nova Micro 1.0Amazon | 35.8% | Not benchmarked | $0.14 | |
| 166 | Mistral LargeMistral AI | 35.1% | Not benchmarked | $6.00 | |
| 167 | Gemma 3 12BGoogle | 34.9% | Not benchmarked | $0.15 | |
| 168 | GPT-4OpenAI | 34.9% | 22.4 | $60.00 | |
| 169 | Granite 4.0 MicroIBM Granite | 33.6% | Not benchmarked | $0.112 | |
| 170 | Mixtral 8x22B InstructMistral AI | 33.2% | Not benchmarked | $6.00 | |
| 171 | GPT-3.5 TurboOpenAI | 29.7% | Not benchmarked | $1.50 | |
| 172 | Gemma 3 4BGoogle | 29.1% | Not benchmarked | $0.05 | |
| 173 | Llama 3.1 8B InstructMeta Llama | 25.9% | Not benchmarked | $0.08 | |
| 174 | Llama 3.2 3B InstructMeta Llama | 25.5% | Not benchmarked | $0.335 | |
| 175 | Llama 3.2 1B InstructMeta Llama | 19.6% | Not benchmarked | $0.201 |
Only models in our catalog are listed, and each appears once with its best published run. 195 rows were read from the source on October 21, 2026, and rows that match no catalog model are left out.