DTBench leaderboard
- Claude Opus 5.5 leads
- 98.9%
- models measured
- 134
- lead over #2
- +0.5 pts
- median
- 83.1%
Leader on DTBench · read October 22
On October 22, 2026, 146 of our catalog's models had a DTBench result, and Anthropic Claude Opus 5.5 led with 98.9%, ahead of Anthropic Claude Fable 5 at 98.4%.
Epoch AI collects its results; its benchmarking hub describes the tasks and links the authors.
Role in the index
DTBench is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 98.9% | 88.8 | $20.00 | |
| 2 | Claude Fable 5Anthropic | 98.4% | 79.3 | $50.00 | |
| 3 | Claude Opus 5Anthropic | 97.9% | 83.0 | $25.00 | |
| 4 | Claude Fable 5.1Anthropic | 97.6% | 84.2 | $50.00 | |
| 5 | Gemini 4 ArgonGoogle | 97.6% | 81.0 | $10.00 | |
| 6 | Claude Sonnet 5.5Anthropic | 97.3% | 87.3 | $10.00 | |
| 7 | GPT-6 AstraOpenAI | 97.3% | 85.1 | $50.00 | |
| 8 | GPT-6 SolOpenAI | 97.3% | 76.0 | $10.00 | |
| 9 | Grok 4.6xAI | 97.3% | 65.7 | $6.00 | |
| 10 | Gemini 3.1 Pro PreviewGoogle | 97.1% | 32.3 | $12.00 | |
| 11 | Gemini 3.7 FlashGoogle | 96.8% | 58.0 | $3.75 | |
| 12 | Muse Spark 1.3Meta | 96.5% | 57.4 | $4.25 | |
| 13 | GPT-6.1 SolOpenAI | 96.5% | 82.4 | $10.00 | |
| 14 | Grok 4.5xAI | 96.5% | 58.6 | $6.00 | |
| 15 | GPT-5.5OpenAI | 96.0% | 71.3 | $30.00 | |
| 16 | GPT-5.5 ProOpenAI | 96.0% | Not benchmarked | $180.00 | |
| 17 | GPT-5.6 Sol ProOpenAI | 96.0% | Not benchmarked | $20.00 | |
| 18 | Grok 4.7xAI | 96.0% | 66.7 | $6.00 | |
| 19 | GPT-5.6 SolOpenAI | 95.7% | 73.5 | $20.00 | |
| 20 | Gemini 3.8 FlashGoogle | 95.7% | 61.6 | $3.75 | |
| 21 | Gemini 3.6 FlashGoogle | 95.5% | 62.5 | $3.75 | |
| 22 | Muse WonderMeta | 95.2% | 71.9 | $20.00 | |
| 23 | Claude Haiku 5.5Anthropic | 95.0% | 70.7 | $0.50 | |
| 24 | Claude Opus 4.8Anthropic | 94.9% | 65.0 | $25.00 | |
| 25 | Gemini 3.5 FlashGoogle | 94.7% | 39.8 | $9.00 | |
| 26 | Muse Spark 1.2Meta | 94.7% | 55.7 | $4.25 | |
| 27 | Claude Opus 4.7Anthropic | 94.7% | 67.5 | $25.00 | |
| 28 | Muse Spark 1.1Meta | 94.4% | 70.3 | $4.25 | |
| 29 | GPT-5.4OpenAI | 94.4% | 64.4 | $15.00 | |
| 30 | DeepSeek V4 Pro 0813DeepSeek | 93.9% | 56.1 | $1.98 | |
| 31 | GLM 5.2Z.ai | 93.6% | 61.1 | $4.40 | |
| 32 | GPT-5.6 TerraOpenAI | 93.3% | 63.9 | $12.00 | |
| 33 | Claude Sonnet 5Anthropic | 92.5% | 57.7 | $10.00 | |
| 34 | Qwen3.7 MaxQwen | 92.3% | 54.1 | $4.425 | |
| 35 | Qwen3.8 Max (0902)Qwen | 92.0% | 62.4 | $6.00 | |
| 36 | Claude Opus 4.6Anthropic | 91.2% | 57.2 | $25.00 | |
| 37 | Kimi K3Moonshot AI | 91.2% | 62.3 | $15.00 | |
| 38 | DeepSeek V4 Flash 0731DeepSeek | 90.9% | 51.8 | $0.396 | |
| 39 | Kimi K2.6Moonshot AI | 90.9% | 52.1 | $4.00 | |
| 40 | GPT-5.2OpenAI | 90.9% | 55.9 | $14.00 | |
| 41 | DeepSeek V4 Pro 0423DeepSeek | 90.7% | 38.7 | $2.976 | |
| 42 | GPT-5OpenAI | 90.7% | 38.8 | $10.00 | |
| 43 | Grok 4.3xAI | 90.7% | 37.4 | $2.50 | |
| 44 | Nemotron 3 UltraNVIDIA | 90.1% | 30.0 | $2.40 | |
| 45 | GPT-5.1OpenAI | 90.1% | 40.4 | $10.00 | |
| 46 | GPT-6 LunaOpenAI | 90.1% | 57.3 | $0.50 | |
| 47 | Grok 4.20xAI | 90.1% | 33.3 | $2.50 | |
| 48 | Claude Opus 4.5Anthropic | 89.9% | 44.0 | $25.00 | |
| 49 | Claude Sonnet 4.6Anthropic | 89.9% | 57.3 | $15.00 | |
| 50 | DeepSeek V4.1 FlashDeepSeek | 89.9% | 62.5 | $0.84 | |
| 51 | Gemini 3 Flash PreviewGoogle | 89.1% | 43.7 | $3.00 | |
| 52 | GPT-5.6 LunaOpenAI | 89.1% | 58.0 | $1.20 | |
| 53 | Mistral Large 4Mistral AI | 88.2% | 56.4 | $4.18 | |
| 54 | Qwen3.8 27BQwen | 88.0% | 74.2 | $2.55 | |
| 55 | DeepSeek V3.2 ExpDeepSeek | 87.7% | Not benchmarked | $0.41 | |
| 56 | GLM 5.3Z.ai | 87.7% | 70.6 | $4.40 | |
| 57 | Qwen3.5 397B A17BQwen | 87.5% | Not benchmarked | $2.34 | |
| 58 | T InklingThinking Machines | 87.5% | 31.5 | $4.05 | |
| 59 | o3 ProOpenAI | 86.9% | Not benchmarked | $80.00 | |
| 60 | DeepSeek V4 Flash 0423DeepSeek | 86.4% | Not benchmarked | $0.28 | |
| 61 | DeepSeek V3.2DeepSeek | 85.6% | 30.0 | $0.88 | |
| 62 | o3OpenAI | 84.8% | Not benchmarked | $8.00 | |
| 63 | MiMo-V2.5-ProXiaomi | 84.5% | 31.6 | $0.9152 | |
| 64 | Qwen3.5-122B-A10BQwen | 84.3% | Not benchmarked | $2.08 | |
| 65 | Qwen3.7 PlusQwen | 84.0% | 33.2 | $1.28 | |
| 66 | Gemini 3.5 Flash LiteGoogle | 83.5% | 49.0 | $2.50 | |
| 67 | Claude Sonnet 4.5Anthropic | 83.2% | 39.5 | $15.00 | |
| 68 | Qwen3.5-FlashQwen | 82.9% | 41.1 | $0.26 | |
| 69 | DeepSeek V3.1DeepSeek | 82.7% | Not benchmarked | $1.50 | |
| 70 | Gemma 4 31BGoogle | 82.7% | 39.0 | $0.40 | |
| 71 | Gemini 2.5 ProGoogle | 82.4% | 19.1 | $10.00 | |
| 72 | Qwen3.5-27BQwen | 82.4% | Not benchmarked | $1.56 | |
| 73 | Qwen3.6 PlusQwen | 81.9% | 44.1 | $1.95 | |
| 74 | Claude Opus 4RetiredAnthropic | 81.6% | Not benchmarked | $75.00 | |
| 75 | DeepSeek V3.1 TerminusDeepSeek | 81.3% | Not benchmarked | $1.0142 | |
| 76 | Qwen-PlusQwen | 81.1% | Not benchmarked | $0.78 | |
| 77 | GPT-5 MiniOpenAI | 80.5% | 34.2 | $2.00 | |
| 78 | Qwen3.5 Plus 2026-04-20Qwen | 80.5% | 39.5 | $1.80 | |
| 79 | GPT-5.4 NanoOpenAI | 80.3% | 44.8 | $1.25 | |
| 80 | Qwen3 235B A22B Thinking 2507Qwen | 80.3% | Not benchmarked | $3.50 | |
| 81 | Claude Opus 4.1Anthropic | 80.0% | Not benchmarked | $75.00 | |
| 82 | GPT-5.4 MiniOpenAI | 80.0% | 35.8 | $4.50 | |
| 83 | Qwen3.5-35B-A3BQwen | 80.0% | Not benchmarked | $1.30 | |
| 84 | MiniMax M3MiniMax | 78.9% | 39.2 | $1.20 | |
| 85 | Qwen3 235B A22B Instruct 2507Qwen | 78.4% | Not benchmarked | $0.75 | |
| 86 | Qwen3.6 27BQwen | 78.1% | 36.8 | $2.70 | |
| 87 | o4 MiniOpenAI | 77.6% | Not benchmarked | $4.40 | |
| 88 | Qwen3.6 FlashQwen | 77.1% | Not benchmarked | $1.125 | |
| 89 | Gemini 3.1 Flash LiteGoogle | 76.8% | Not benchmarked | $1.50 | |
| 90 | Gemini 2.5 FlashGoogle | 76.5% | Not benchmarked | $2.50 | |
| 91 | gpt-oss-120bOpenAI | 76.3% | Not benchmarked | $0.50 | |
| 92 | Mistral Medium 3.5Mistral AI | 75.5% | 25.5 | $7.50 | |
| 93 | Gemma 4 26B A4BGoogle | 74.9% | Not benchmarked | $0.60 | |
| 94 | o1OpenAI | 74.7% | Not benchmarked | $60.00 | |
| 95 | Qwen3.6 35B A3BQwen | 73.9% | Not benchmarked | $1.00 | |
| 96 | Claude Haiku 4.5Anthropic | 73.6% | 35.6 | $5.00 | |
| 97 | Qwen3.5-9BQwen | 71.2% | Not benchmarked | $0.15 | |
| 98 | Mistral Small 4Mistral AI | 70.9% | Not benchmarked | $0.60 | |
| 99 | GPT-4.1 MiniOpenAI | 68.8% | Not benchmarked | $1.60 | |
| 100 | o3 MiniOpenAI | 68.8% | Not benchmarked | $4.40 | |
| 101 | GPT-4.1OpenAI | 68.3% | Not benchmarked | $8.00 | |
| 102 | gpt-oss-20bOpenAI | 68.0% | Not benchmarked | $0.15 | |
| 103 | Claude 3.5 SonnetRetiredAnthropic | 67.8% | Not benchmarked | $15.00 | |
| 104 | Mistral Medium 3.1Mistral AI | 67.5% | Not benchmarked | $2.00 | |
| 105 | Qwen3 32BQwen | 67.5% | Not benchmarked | $0.40 | |
| 106 | Qwen3 30B A3B Instruct 2507Qwen | 67.2% | Not benchmarked | $0.30 | |
| 107 | Grok 2RetiredxAI | 65.2% | Not benchmarked | $10.00 | |
| 108 | Mistral Large 3 2512Mistral AI | 65.1% | Not benchmarked | $1.50 | |
| 109 | DeepSeek V3 0324DeepSeek | 64.8% | Not benchmarked | $1.07 | |
| 110 | GPT-4oOpenAI | 64.5% | Not benchmarked | $10.00 | |
| 111 | Qwen3 14BQwen | 64.0% | Not benchmarked | $0.23 | |
| 112 | Gemini 2.0 FlashRetiredGoogle | 63.2% | Not benchmarked | $0.40 | |
| 113 | Qwen2.5 72B InstructQwen | 62.9% | Not benchmarked | $0.40 | |
| 114 | Gemini 2.5 Flash LiteGoogle | 62.8% | Not benchmarked | $0.40 | |
| 115 | DeepSeek V3DeepSeek | 62.7% | Not benchmarked | $0.9594 | |
| 116 | GPT-4OpenAI | 62.7% | Not benchmarked | $60.00 | |
| 117 | GPT-5 NanoOpenAI | 62.7% | Not benchmarked | $0.40 | |
| 118 | Mistral Medium 3Mistral AI | 62.3% | Not benchmarked | $2.00 | |
| 119 | Llama 4 MaverickMeta Llama | 61.9% | Not benchmarked | $1.00 | |
| 120 | Claude 3 OpusRetiredAnthropic | 61.6% | Not benchmarked | $75.00 | |
| 121 | GPT-4 TurboOpenAI | 61.6% | Not benchmarked | $30.00 | |
| 122 | Command ACohere | 61.3% | Not benchmarked | $10.00 | |
| 123 | Mistral Large 2407Mistral AI | 61.2% | Not benchmarked | $6.00 | |
| 124 | Mistral LargeMistral AI | 60.8% | Not benchmarked | $6.00 | |
| 125 | Qwen3 30B A3BQwen | 60.3% | Not benchmarked | $0.50 | |
| 126 | Llama 3.1 70B InstructMeta Llama | 60.0% | Not benchmarked | $0.56 | |
| 127 | Llama 3.3 70B InstructMeta Llama | 59.5% | Not benchmarked | $0.72 | |
| 128 | Gemini 1.5 ProRetiredGoogle | 59.0% | Not benchmarked | $5.00 | |
| 129 | Llama 4 ScoutMeta Llama | 57.9% | Not benchmarked | $0.50 | |
| 130 | Claude 3.5 HaikuRetiredAnthropic | 56.7% | Not benchmarked | $4.00 | |
| 131 | Mixtral 8x22B InstructMistral AI | 55.1% | Not benchmarked | $6.00 | |
| 132 | GPT-4o-miniOpenAI | 54.4% | Not benchmarked | $0.60 | |
| 133 | Gemini 1.5 FlashRetiredGoogle | 53.8% | Not benchmarked | $0.30 | |
| 134 | Claude 3 SonnetRetiredAnthropic | 53.6% | Not benchmarked | $15.00 | |
| 135 | Gemini 2.0 Flash-LiteRetiredGoogle | 52.5% | Not benchmarked | $0.30 | |
| 136 | Gemma 3 27BGoogle | 52.5% | Not benchmarked | $0.30 | |
| 137 | GPT-4.1 NanoOpenAI | 52.5% | Not benchmarked | $0.40 | |
| 138 | Claude 2RetiredAnthropic | 51.9% | Not benchmarked | $24.00 | |
| 139 | Gemma 3 4BGoogle | 50.9% | Not benchmarked | $0.075 | |
| 140 | Llama 3.1 8B InstructMeta Llama | 50.9% | Not benchmarked | $0.08 | |
| 141 | Claude 3 HaikuRetiredAnthropic | 50.1% | Not benchmarked | $1.25 | |
| 142 | Gemma 3 12BGoogle | 48.8% | Not benchmarked | $0.1625 | |
| 143 | Mistral NemoMistral AI | 48.6% | Not benchmarked | $0.15 | |
| 144 | GPT-3.5 TurboOpenAI | 48.5% | Not benchmarked | $1.50 | |
| 145 | Gemma 2 27BGoogle | 48.0% | Not benchmarked | $0.65 | |
| 146 | Qwen2.5 7B InstructQwen | 47.7% | Not benchmarked | $0.20 |
Only models in our catalog are listed, and each appears once with its best published run. 215 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out.