Terminal-Bench 2.0
Hard tasks done entirely in a Linux terminal: building, debugging, data work, system administration. Best published agent per model.
| Category | Agents |
|---|---|
| Metric | Accuracy |
| Source | Epoch AI (evaluator) |
| License | CC BY 4.0 |
| Benchmark page | www.tbench.ai |
| Status | Scored |
| Weight | 1.2 within Agents |
| Fit | Discrimination 0.89, noise 0.28 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | GPT-5.5 | OpenAI | 84.7% | 2026-04-23 |
| 2 | GPT-5.5 | OpenAI | 83.2% | 2026-04-23 |
| 3 | GPT-5.5 | OpenAI | 82.3% | 2026-04-23 |
| 4 | GPT-5.5 | OpenAI | 82.0% | 2026-04-23 |
| 5 | GPT-5.4 | OpenAI | 81.8% | 2026-03-12 |
| 6 | Claude Opus 4.7 | Anthropic | 80.2% | 2026-04-16 |
| 7 | Gemini 3.1 Pro Preview | 80.2% | 2026-02-19 | |
| 8 | Claude Opus 4.6 | Anthropic | 79.8% | 2026-03-12 |
| 9 | GPT-5.3 Codex | OpenAI | 78.4% | 2026-02-05 |
| 10 | Gemini 3.1 Pro Preview | 78.4% | 2026-03-02 | |
| 11 | Gemini 3.1 Pro Preview | 78.4% | 2026-03-02 | |
| 12 | GPT-5.3 Codex | OpenAI | 77.3% | 2026-02-05 |
| 13 | Claude Opus 4.6 | Anthropic | 76.4% | 2026-02-05 |
| 14 | GPT-5.3 Codex | OpenAI | 75.8% | 2026-02-05 |
| 15 | GPT-5.3 Codex | OpenAI | 75.7% | 2026-02-05 |
| 16 | Claude Opus 4.6 | Anthropic | 75.3% | 2026-02-05 |
| 17 | GPT-5.3 Codex | OpenAI | 75.1% | 2026-02-05 |
| 18 | Gemini 3.1 Pro Preview | 74.8% | 2026-02-19 | |
| 19 | Claude Opus 4.6 | Anthropic | 74.7% | 2026-02-05 |
| 20 | GPT-5.3 Codex | OpenAI | 74.6% | 2026-02-05 |
| 21 | Claude Opus 4.6 | Anthropic | 72.1% | 2026-02-05 |
| 22 | Claude Opus 4.6 | Anthropic | 71.9% | 2026-02-05 |
| 23 | GPT-5.3 Codex | OpenAI | 71.5% | 2026-02-05 |
| 24 | GPT-5.3 Codex | OpenAI | 70.3% | 2026-02-10 |
| 25 | Claude Opus 4.6 | Anthropic | 69.9% | 2026-02-05 |
| 26 | Gemini 3 Pro Preview | 69.4% | 2025-11-18 | |
| 27 | GPT-5.3 Codex | OpenAI | 69.1% | 2026-02-05 |
| 28 | Claude Opus 4.6 | Anthropic | 66.9% | 2026-02-05 |
| 29 | Claude Opus 4.6 | Anthropic | 66.5% | 2026-02-05 |
| 30 | GPT-5.2 Codex | OpenAI | 66.5% | 2026-01-14 |
| 31 | GPT-5.5 | OpenAI | 66.1% | 2026-04-23 |
| 32 | Gemini 3 Pro Preview | 65.2% | 2025-11-18 | |
| 33 | GPT-5.2 | OpenAI | 64.9% | 2025-12-11 |
| 34 | GPT-5.2 Medium | OpenAI | 64.9% | 2025-12-24 |
| 35 | GPT-5.3 Codex | OpenAI | 64.7% | 2026-02-05 |
| 36 | Gemini 3 Flash Preview | 64.3% | 2025-12-17 | |
| 37 | Claude Opus 4.5 | Anthropic | 63.2% | 2025-11-24 |
| 38 | Claude Opus 4.6 | Anthropic | 62.9% | 2026-02-05 |
| 39 | GPT-5.2 | OpenAI | 62.9% | 2025-12-11 |
| 40 | GPT-5.2 Medium | OpenAI | 62.9% | 2025-12-18 |
| 41 | Gemini 3 Pro Preview | 62.3% | 2025-11-18 | |
| 42 | Gemini 3 Pro Preview | 62.2% | 2026-02-05 | |
| 43 | Gemini 3 Pro Preview | 61.8% | 2025-11-18 | |
| 44 | GPT-5.1 Codex mini | OpenAI | 61.6% | 2025-11-12 |
| 45 | Gemini 3.1 Pro Preview | 61.4% | 2026-02-19 | |
| 46 | Gemini 3 Pro Preview | 61.1% | 2025-11-18 | |
| 47 | GPT-5.2 | OpenAI | 60.7% | 2025-12-11 |
| 48 | GPT-5.1 Codex Max | OpenAI | 60.5% | 2025-11-19 |
| 49 | Claude Opus 4.5 | Anthropic | 59.1% | 2025-11-24 |
| 50 | Claude Opus 4.5 Thinking | Anthropic | 59.1% | 2025-12-17 |
| 51 | Claude Opus 4.5 | Anthropic | 58.4% | 2025-11-24 |
| 52 | Claude Opus 4.6 | Anthropic | 58.0% | 2026-02-05 |
| 53 | GPT-5.1 Codex | OpenAI | 57.8% | 2025-11-16 |
| 54 | Claude Opus 4.5 | Anthropic | 57.8% | 2025-11-24 |
| 55 | GPT-5.1 Codex | OpenAI | 57.8% | 2025-11-12 |
| 56 | Grok 4.20 (Reasoning) | xAI | 57.3% | 2026-03-10 |
| 57 | Grok 4.20 (Reasoning) | xAI | 57.3% | 2026-04-02 |
| 58 | Gemini 3 Pro Preview | 56.9% | 2025-11-18 | |
| 59 | Gemini 3 Pro Preview | 56.0% | 2025-11-18 | |
| 60 | Claude Opus 4.5 | Anthropic | 54.3% | 2025-11-24 |
| 61 | GPT-5.2 | OpenAI | 54.0% | 2025-12-11 |
| 62 | GPT-5.1 Codex | OpenAI | 53.5% | 2025-11-12 |
| 63 | Claude Sonnet 4.6 | Anthropic | 53.4% | 2026-02-17 |
| 64 | GLM-5 | Zhipu AI | 52.4% | 2026-02-11 |
| 65 | Claude Opus 4.5 | Anthropic | 52.1% | 2025-11-24 |
| 66 | Claude Opus 4.5 | Anthropic | 51.9% | 2025-11-24 |
| 67 | Claude Opus 4.5 | Anthropic | 51.7% | 2025-11-24 |
| 68 | Gemini 3 Flash Preview | 51.7% | 2025-12-17 | |
| 69 | Gemini 3 Flash Preview | 51.0% | 2025-12-23 | |
| 70 | GPT-5 | OpenAI | 49.6% | 2025-08-07 |
| 71 | GPT-5 Medium | OpenAI | 49.6% | 2025-11-04 |
| 72 | GPT-5.1 Medium | OpenAI | 47.6% | 2025-11-16 |
| 73 | GPT-5.1 | OpenAI | 47.6% | 2025-11-12 |
| 74 | Claude Sonnet 4.5 | Anthropic | 46.5% | 2025-09-29 |
| 75 | MiniMax-M2.7 | MiniMax | 45.1% | 2026-03-18 |
| 76 | GPT-5-Codex | OpenAI | 44.3% | 2025-09-15 |
| 77 | GPT-5 | OpenAI | 43.8% | 2025-08-07 |
| 78 | GPT-5 Medium | OpenAI | 43.8% | 2025-11-02 |
| 79 | GPT-5-Codex | OpenAI | 43.4% | 2025-09-15 |
| 80 | Kimi K2.5 | Moonshot AI | 43.2% | 2026-01-27 |
| 81 | Claude Sonnet 4.5 | Anthropic | 43.2% | 2025-09-29 |
| 82 | GPT-5.1 Codex mini | OpenAI | 43.2% | 2025-11-12 |
| 83 | GPT-5.1 Codex mini | OpenAI | 43.1% | 2025-11-17 |
| 84 | MiniMax-M2.7 | MiniMax | 42.9% | 2026-03-18 |
| 85 | Claude Sonnet 4.5 | Anthropic | 42.8% | 2025-09-29 |
| 86 | Claude Sonnet 4.5 | Anthropic | 42.7% | 2026-01-04 |
| 87 | Claude Sonnet 4.5 | Anthropic | 42.7% | 2025-09-29 |
| 88 | MiniMax-M2.5 | MiniMax | 42.7% | 2026-02-12 |
| 89 | Claude Sonnet 4.5 | Anthropic | 42.6% | 2025-11-02 |
| 90 | Claude Sonnet 4.5 | Anthropic | 42.5% | 2025-09-29 |
| 91 | MiniMax-M2.5 | MiniMax | 42.3% | 2026-02-12 |
| 92 | GPT-5-Codex | OpenAI | 41.4% | 2025-09-15 |
| 93 | Claude Sonnet 4.5 | Anthropic | 40.1% | 2025-11-04 |
| 94 | DeepSeek V3.2 | DeepSeek | 39.6% | 2025-12-01 |
| 95 | Claude Opus 4.1 | Anthropic | 38.0% | 2025-10-31 |
| 96 | Claude Opus 4.1 | Anthropic | 36.9% | 2025-11-02 |
| 97 | GPT-5.1 Codex | OpenAI | 36.9% | 2025-11-17 |
| 98 | MiniMax-M2.1 | MiniMax | 36.6% | 2025-12-23 |
| 99 | Kimi K2 Thinking | Moonshot AI | 35.7% | 2025-11-06 |
| 100 | Claude Haiku 4.5 | Anthropic | 35.5% | 2025-10-15 |
| 101 | GPT-5 Medium | OpenAI | 35.2% | 2025-10-31 |
| 102 | GPT-5 | OpenAI | 35.2% | 2025-08-07 |
| 103 | Claude Opus 4.1 | Anthropic | 35.1% | 2025-11-03 |
| 104 | Claude Opus 4.1 | Anthropic | 34.8% | 2025-08-05 |
| 105 | GPT-5 Mini | OpenAI | 34.8% | 2025-08-07 |
| 106 | GPT-5 Mini | OpenAI | 34.8% | 2025-12-24 |
| 107 | GPT-5 | OpenAI | 33.9% | 2025-08-07 |
| 108 | GPT-5 Medium | OpenAI | 33.9% | 2025-11-03 |
| 109 | GLM-4.7 | Zhipu AI | 33.4% | 2025-12-22 |
| 110 | GLM-4.7 | Zhipu AI | 33.3% | 2025-12-22 |
| 111 | Gemini 2.5 Pro | 32.6% | 2025-06-17 | |
| 112 | GPT-5 Mini Medium | OpenAI | 31.9% | 2025-11-04 |
| 113 | GPT-5 Mini | OpenAI | 31.9% | 2025-08-07 |
| 114 | MiniMax-M2 | MiniMax | 30.0% | 2025-10-27 |
| 115 | Claude Haiku 4.5 | Anthropic | 29.8% | 2025-10-15 |
| 116 | MiniMax-M2.1 | MiniMax | 29.2% | 2025-12-23 |
| 117 | GPT-5 Mini | OpenAI | 29.2% | 2025-08-07 |
| 118 | GPT-5 Mini Medium | OpenAI | 29.2% | 2025-11-02 |
| 119 | Claude Haiku 4.5 | Anthropic | 28.3% | 2025-10-15 |
| 120 | Kimi K2 Instruct | Moonshot AI | 27.8% | 2025-11-01 |
| 121 | Claude Haiku 4.5 | Anthropic | 27.5% | 2025-10-15 |
| 122 | Grok 4 | xAI | 27.2% | 2025-11-02 |
| 123 | Qwen3-Coder 480B-A35B Instruct | Alibaba | 27.2% | 2025-07-22 |
| 124 | Kimi K2 Instruct | Moonshot AI | 26.7% | 2025-07-11 |
| 125 | Gemini 2.5 Pro | 26.1% | 2025-11-03 | |
| 126 | x-AI/Grok-Code-Fast 1 | xAI | 25.8% | 2025-08-28 |
| 127 | Qwen3-Coder 480B-A35B Instruct | Alibaba | 25.4% | 2025-11-02 |
| 128 | Grok 4 | xAI | 25.4% | 2025-11-03 |
| 129 | Qwen3.6 35B-A3B | Alibaba | 24.6% | 2026-04-16 |
| 130 | GLM-4.6 | Zhipu AI | 24.5% | 2025-09-30 |
| 131 | GPT-5 Mini | OpenAI | 24.0% | 2025-08-07 |
| 132 | GPT-5 Mini Medium | OpenAI | 24.0% | 2025-10-31 |
| 133 | Qwen3-Coder 480B-A35B Instruct | Alibaba | 23.9% | 2025-11-01 |
| 134 | Grok 4 | xAI | 23.2% | 2025-07-09 |
| 135 | GPT-5 Mini | OpenAI | 22.3% | 2025-08-07 |
| 136 | GPT-5 Mini Medium | OpenAI | 22.2% | 2025-11-03 |
| 137 | GPT-5 Nano | OpenAI | 21.8% | 2025-08-07 |
| 138 | Gemini 2.5 Pro | 19.6% | 2025-11-04 | |
| 139 | GPT OSS 120B | OpenAI | 18.7% | 2025-11-01 |
| 140 | Gemini 2.5 Flash (Sep 2025) | 17.1% | 2025-11-03 | |
| 141 | Gemini 2.5 Flash | 17.1% | 2025-06-17 | |
| 142 | Gemini 2.5 Flash (Sep 2025) | 16.9% | 2025-10-31 | |
| 143 | Gemini 2.5 Flash | 16.9% | 2025-06-17 | |
| 144 | Gemini 2.5 Pro | 16.4% | 2025-06-17 | |
| 145 | Gemini 2.5 Flash | 16.4% | 2025-06-17 | |
| 146 | Gemini 2.5 Flash (Sep 2025) | 16.4% | 2025-11-02 | |
| 147 | Gemini 2.5 Flash | 15.5% | 2025-06-17 | |
| 148 | Gemini 2.5 Flash (Sep 2025) | 15.4% | 2025-11-04 | |
| 149 | x-AI/Grok-Code-Fast 1 | xAI | 14.2% | 2025-10-31 |
| 150 | GPT OSS 120B | OpenAI | 14.2% | 2025-11-03 |
| 151 | Claude Haiku 4.5 | Anthropic | 13.9% | 2025-10-15 |
| 152 | GPT-5 Nano Medium | OpenAI | 11.5% | 2025-11-04 |
| 153 | GPT-5 Nano | OpenAI | 11.5% | 2025-08-07 |
| 154 | GPT-5 Nano Medium | OpenAI | 9.9% | 2025-11-02 |
| 155 | GPT-5 Nano | OpenAI | 9.9% | 2025-08-07 |
| 156 | Qwen3.5 9B | Alibaba | 9.2% | 2026-03-02 |
| 157 | GPT-5 Nano Medium | OpenAI | 7.9% | 2025-10-31 |
| 158 | GPT-5 Nano | OpenAI | 7.9% | 2025-08-07 |
| 159 | GPT-5 Nano Medium | OpenAI | 7.0% | 2025-11-03 |
| 160 | GPT-5 Nano | OpenAI | 7.0% | 2025-08-07 |
| 161 | GPT OSS 20B | OpenAI | 3.4% | 2025-11-03 |
| 162 | GPT OSS 20B | OpenAI | 3.1% | 2025-11-01 |