Terminal-Bench 2.1 (Vals run)
Difficult terminal tasks from Terminal-Bench 2.1, a revision of 2.0. Shares weight with Terminal-Bench 2.0. Run by Vals AI.
| Category | Agents |
|---|---|
| Metric | Accuracy (%) |
| Source | Vals AI (evaluator) |
| License | None stated |
| Benchmark page | www.vals.ai |
| Status | Scored |
| Weight | 0.9 within Agents |
| Fit | Discrimination 0.80, noise 0.29 |
| Lineage | Shares weight with terminal_bench_2 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 High | Anthropic | 87.6% | 2026-10-01 |
| 2 | GPT-6 Astra Max | OpenAI | 87.3% | 2026-10-01 |
| 3 | GPT-5.6 Sol Max | OpenAI | 85.8% | 2026-10-01 |
| 4 | Claude Fable 5.1 High | Anthropic | 85.0% | 2026-10-01 |
| 5 | Claude Opus 5 High | Anthropic | 84.6% | 2026-10-01 |
| 6 | GPT-6 Sol Max | OpenAI | 83.1% | 2026-10-01 |
| 7 | Claude Sonnet 5.5 High | Anthropic | 83.1% | 2026-10-01 |
| 8 | Gemini 3.8 Flash High | 81.3% | 2026-10-01 | |
| 9 | Kimi K3 | Moonshot AI | 80.9% | 2026-10-01 |
| 10 | Claude Fable 5 Max | Anthropic | 80.5% | 2026-10-01 |
| 11 | GPT-5.6 Luna Max | OpenAI | 79.0% | 2026-10-01 |
| 12 | Muse Spark 1.3 Max | Meta | 79.0% | 2026-10-01 |
| 13 | Grok 4.6 High | xAI | 78.3% | 2026-10-01 |
| 14 | Gemini 3.7 Flash High | 77.5% | 2026-10-01 | |
| 15 | GPT-5.6 Terra Max | OpenAI | 77.5% | 2026-10-01 |
| 16 | MiMo-V2.6-Flash | Xiaomi | 76.4% | 2026-10-01 |
| 17 | GPT-5.5 High | OpenAI | 76.4% | 2026-10-01 |
| 18 | Claude Sonnet 5 Max | Anthropic | 74.5% | 2026-10-01 |
| 19 | DeepSeek V4.1 Flash High | DeepSeek | 74.5% | 2026-10-01 |
| 20 | Gemini 3.5 Flash High | 74.2% | 2026-10-01 | |
| 21 | Gemini 3.6 Flash High | 73.8% | 2026-10-01 | |
| 22 | Grok 4.7 Extra High | xAI | 73.4% | 2026-10-01 |
| 23 | GPT-6 Luna Max | OpenAI | 73.0% | 2026-10-01 |
| 24 | Muse Spark 1.3 Extra High | Meta | 72.3% | 2026-10-01 |
| 25 | Claude Opus 4.8 Max | Anthropic | 71.9% | 2026-10-01 |
| 26 | GLM-5.3 Max | Zhipu AI | 71.5% | 2026-10-01 |
| 27 | Gemini 3.1 Pro Preview High | 70.8% | 2026-10-01 | |
| 28 | Claude Opus 4.8 | Anthropic | 69.7% | 2026-10-01 |
| 29 | Muse Spark 1.2 Extra High | Meta | 69.7% | 2026-10-01 |
| 30 | Muse Spark 1.1 Extra High | Meta | 69.3% | 2026-10-01 |
| 31 | Claude Opus 4.7 High | Anthropic | 68.5% | 2026-10-01 |
| 32 | MiMo-V2.6-Pro | Xiaomi | 67.8% | 2026-10-01 |
| 33 | Grok 4.5 High | xAI | 67.8% | 2026-10-01 |
| 34 | GLM-5.2 Max | Zhipu AI | 67.8% | 2026-10-01 |
| 35 | Qwen3.8 Max | Alibaba | 67.4% | 2026-10-01 |
| 36 | Kimi K2.7 Code | Moonshot AI | 67.0% | 2026-10-01 |
| 37 | DeepSeek V4 Flash 0731 | DeepSeek | 67.0% | 2026-10-01 |
| 38 | GLM-5.3-Flash Max | Zhipu AI | 62.9% | 2026-10-01 |
| 39 | Qwen3.7 Max | Alibaba | 61.0% | 2026-10-01 |
| 40 | MiMo-V2.5 | Xiaomi | 60.7% | 2026-10-01 |
| 41 | Composer 2.5 | Cursor | 58.4% | 2026-10-01 |
| 42 | Qwen3.8 27B Extra High | Alibaba | 58.4% | 2026-10-01 |
| 43 | Gemini 4 Argon High | 57.7% | 2026-10-01 | |
| 44 | MiMo-V2.5-Pro | Xiaomi | 57.3% | 2026-10-01 |
| 45 | Claude Sonnet 4.6 Max | Anthropic | 57.3% | 2026-10-01 |
| 46 | GLM-5.1 | Zhipu AI | 56.9% | 2026-10-01 |
| 47 | Hy4 preview | Tencent | 55.1% | 2026-10-01 |
| 48 | Inkling Small | Thinking Machines | 55.1% | 2026-10-01 |
| 49 | DeepSeek V4 Pro 0813 Max | DeepSeek | 54.7% | 2026-10-01 |
| 50 | GPT-5.4 mini High | OpenAI | 54.7% | 2026-10-01 |
| 51 | Gemini 3 Flash Preview High | 53.9% | 2026-10-01 | |
| 52 | Kimi K2.6 | Moonshot AI | 53.6% | 2026-10-01 |
| 53 | MiniMax-M3 | MiniMax | 53.6% | 2026-10-01 |
| 54 | Qwen3.6 Plus | Alibaba | 53.2% | 2026-10-01 |
| 55 | Qwen3.7 Plus | Alibaba | 52.8% | 2026-10-01 |
| 56 | Nemotron 3 Ultra 550B A55B | NVIDIA | 50.9% | 2026-10-01 |
| 57 | DeepSeek V4 Pro | DeepSeek | 50.2% | 2026-10-01 |
| 58 | Gemini 3.5 Flash Lite High | 50.2% | 2026-10-01 | |
| 59 | MiniMax-M2.7 | MiniMax | 48.7% | 2026-10-01 |
| 60 | Inkling | Thinking Machines | 47.6% | 2026-10-01 |
| 61 | Grok 4.20 (Reasoning) | xAI | 44.2% | 2026-10-01 |
| 62 | Claude Haiku 4.5 Thinking | Anthropic | 43.8% | 2026-10-01 |
| 63 | Grok 4.3 High | xAI | 41.9% | 2026-10-01 |
| 64 | Kimi K2.5 Thinking | Moonshot AI | 41.9% | 2026-10-01 |
| 65 | GPT-5.4 nano High | OpenAI | 41.6% | 2026-10-01 |
| 66 | Mistral Medium 3.5 High | Mistral | 39.0% | 2026-10-01 |
| 67 | Mercury 2.5 High | Inception | 34.5% | 2026-10-01 |
| 68 | Laguna M.1 | Poolside | 34.1% | 2026-10-01 |
| 69 | Gemini 3.1 Flash Lite Preview High | 34.1% | 2026-10-01 | |
| 70 | Laguna XS.2 | Poolside | 25.8% | 2026-10-01 |
| 71 | Command A Plus | Cohere | 17.6% | 2026-10-01 |