Inference
EN
Vals AI

Terminal-Bench 2.1

State-of-the-art set of difficult terminal-based tasks

ModelsModels with a result.75
Top resultBest result on this benchmark.87.6%Claude Opus 5.5
Top-3 spreadPoints from first to third.1.9 pts
UpdatedDate the source changed the results.1 Oct 2026

Result and cost

72 of 75 results

Results

75 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Claude Opus 5.587.6% ±1.7$0.4966 min
02GPT-6 Astra87.3% ±0.4$1.345 min
03GPT-5.6 Sol85.8% ±1.4$1.026 min
04Claude Fable 5.185.0% ±1.0$3.078 min
05Claude Opus 584.6% ±1.0$0.88713 min
06Claude Sonnet 5.583.1% ±1.7$0.6258 min
06GPT-6 Sol83.1% ±1.3$0.3735 min
08Gemini 3.8 Flash81.3% ±0.4$1.549 min
09Kimi K380.9% ±0.6$0.34110 min
10Claude Fable 580.5% ±1.4$1.438 min
11GPT-5.6 Luna79.0% ±1.0$0.0546 min
11Muse Spark 1.3 Max79.0% ±1.0$0.59712 min
13Grok 4.678.3% ±2.1$0.45411 min
14GPT-5.6 Terra77.5% ±2.2$0.4739 min
14Gemini 3.7 Flash77.5% ±0.6$1.028 min
16GPT-5.576.4% ±0.6$0.7427 min
16MiMo-V2.6-Flash76.4% ±1.7$0.02313 min
18DeepSeek V4.1 Flash 101774.5% ±1.6$0.09910 min
18Claude Sonnet 574.5% ±2.1$0.53511 min
20Gemini 3.5 Flash74.2% ±1.1$0.8686 min
21Gemini 3.6 Flash73.8% ±1.6$1.179 min
22Grok 4.773.4% ±1.5$1.0813 min
23GPT-6 Luna73.0% ±1.7$0.0297 min
24Muse Spark 1.372.3% ±0.4$0.66314 min
25Claude Opus 4.871.9% ±0.6$2.4115 min
26GLM-5.371.5% ±2.0$0.30713 min
27Gemini 3.1 Pro Preview70.8% ±1.1$0.5837 min
28Muse Spark 1.269.7% ±0.6$0.49915 min
28Claude Opus 4.8 Claude CodeClaude Code69.7% ±4.9$0.9686 min
30Muse Spark 1.169.3% ±1.0$0.41710 min
31Claude Opus 4.768.5% ±1.3$1.9814 min
32MiMo-V2.6-Pro67.8% ±1.9$0.07316 min
32Grok 4.567.8% ±4.4$0.28410 min
32GLM-5.267.8% ±1.0$0.42514 min
35Qwen3.8 Max67.4% ±0.6$0.44815 min
36DeepSeek V4 Flash 073167.0% ±1.4$0.02111 min
36Kimi K2.7 Code67.0% ±0.4$0.25713 min
38GLM-5.3-Flash62.9% ±2.2$0.02816 min
39Qwen3.7 Max61.0% ±0.4$0.33216 min
40MiMo-V2.560.7% ±5.2$0.02213 min
41Qwen3.8-27B58.4% ±5.3$0.67815 min
41Composer 2.5Cursor CLI58.4% ±5.3—0.0 s
43Gemini 4 Argon57.7% ±2.5—14 min
44Claude Sonnet 4.657.3% ±0.6$0.57212 min
44MiMo-V2.5-Pro57.3% ±5.3$0.04414 min
44GPT-5.5 CodexCodex57.3% ±5.3$0.6453 min
44GPT-5.5 FactoryFactory57.3% ±5.3$0.7526 min
48GLM-5.156.9% ±1.4$0.23216 min
49Hy4 preview55.1% ±3.4$0.1418 min
49Inkling-Small55.1% ±2.3$0.12810 min
51DeepSeek V4 Pro 081354.7% ±1.5$0.2014 min
51GPT-5.4 mini54.7% ±4.6$0.28711 min
53Gemini 3 Flash Preview53.9% ±1.3$0.1546 min
54Kimi K2.653.6% ±2.1$0.13816 min
54MiniMax M353.6% ±0.7$0.20617 min
56Qwen3.6 Plus53.2% ±1.6$0.20213 min
57Qwen3.7 Plus52.8% ±0.6$0.09611 min
58DeepSeek V4 Pro 042350.2% ±1.5$0.23918 min
58Gemini 3.5 Flash-Lite50.2% ±1.0$0.3377 min
58Ling 3.0 Flash50.2% ±2.5$0.06710 min
58Ling 3.0 Flash Af Rc350.2% ±1.4$0.04314 min
62MiniMax M2.748.7% ±3.1$0.06614 min
63Inkling47.6% ±0.4$0.63113 min
64Grok 4.20 (Reasoning)44.2% ±1.0$0.4075 min
65Claude Haiku 4.5 Thinking43.8% ±1.9$0.36411 min
66Grok 4.341.9% ±2.6$0.9828 min
66Kimi K2.5 Thinking41.9% ±0.7$0.08618 min
68GPT-5.4 nano41.6% ±3.0$0.07213 min
69Mistral Medium 3.539.0% ±2.7$1.1916 min
70Mercury 2.534.5% ±0.7$0.21511 min
71Gemini 3.1 Flash Lite Preview34.1% ±1.0$0.0855 min
71Laguna M.134.1% ±1.6$0.30120 min
73Laguna XS.225.8% ±3.0$0.25420 min
74Command A+17.6% ±0.4—0.0 s
75Nemotron Lightning 3p5 30B A3B10.9% ±2.6$0.01417 min