Inference
EN
Vals AI

Terminal-Bench 4.0

Frontier-difficulty terminal tasks across software, science, ML, operations, hardware, security, and media

ModelsModels with a result.61
Top resultBest result on this benchmark.65.2%Claude Fable 5.2
Top-3 spreadPoints from first to third.1.0 pts
UpdatedDate the source changed the results.9 Nov 2026

Result and cost

60 of 61 results

Results

61 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01Claude Fable 5.265.2% ±2.3$14.5655 min
01Claude Opus 5.565.2%$12.0464 min
03Claude Sonnet 5.564.1% ±1.0$10.7582 min
04Gemini 4 Carbon63.1% ±2.2$7.4545 min
05GPT-6 Astra59.6% ±4.4$9.5836 min
06Muse Wonder Full58.1% ±3.3$10.5372 min
06Claude Fable 5.158.1% ±3.3$17.1860 min
08Gemini 4 Argon57.6% ±2.3$17.6454 min
09GPT-6.1 Sol55.1% ±1.8$1.7245 min
10Claude Opus 553.5% ±1.3$18.6067 min
11GPT-6.1 Terra49.5% ±1.8$2.1634 min
11Muse Wonder49.5% ±2.2$14.6655 min
13GPT-6 Sol44.4% ±3.5$5.7935 min
14Claude Fable 541.4% ±1.3$30.3468 min
15Hy439.4% ±0.9$1.2843 min
15Muse Ember39.4% ±2.3$0.21826 min
17GLM-5.338.9% ±1.8$9.3783 min
17Qwen3.8 Max Internal38.9% ±1.3$12.152.4 h
19GPT-5.6 Sol37.9%$7.9834 min
20GLM-5.3-Flash37.4% ±2.7$1.2261 min
21Claude Haiku 5.536.4% ±2.3$3.1252 min
22Qwen3.8 Max34.3% ±3.9$10.632.5 h
23Step 5 Preview31.8% ±2.6$3.112.0 h
24MiMo-V2.6-Pro31.3% ±3.1$0.4572.7 h
25Grok 4.728.8% ±2.3$18.0150 min
26Mistral Large 427.8% ±1.8$5.181.7 h
27Mistral Large 4 OSS26.3% ±5.1$2.7365 min
28MiniMax M3.1 Flash25.3% ±1.0$0.73551 min
29Muse Spark 1.3 Max24.7% ±0.5$6.6553 min
30MiMo-V2.6-Flash24.2% ±1.5$0.2122.4 h
31Claude Opus 4.823.2% ±1.3$17.1471 min
32GPT-5.6 Terra22.7% ±2.3$5.6031 min
33Ember-120.7% ±3.6$19.481.9 h
34DeepSeek V4.1 Flash 101719.7% ±1.8$0.49948 min
34Gemini 3.8 Flash19.7% ±2.3$8.651.8 h
36DeepSeek V4 Flash 073118.7% ±3.5$0.7661.8 h
37Grok 4.617.2% ±1.3$5.0629 min
37Kimi K317.2% ±0.5$12.483.2 h
39DeepSeek V4 Pro 081314.1% ±2.0$3.311.5 h
40GPT-6 Luna13.6% ±1.5$0.35533 min
41Inkling 2 Small12.6% ±3.1$8.221.7 h
42Gemini 3.7 Flash12.1% ±3.0$8.7184 min
43GPT-5.6 Luna11.6% ±1.0$0.73236 min
44DeepSeek V4 Pro 042311.1% ±1.3$3.4586 min
45Muse Spark 1.310.6% ±1.8$7.042.0 h
45Claude Sonnet 510.6% ±1.5$26.051.7 h
47Grok 4.58.6% ±1.3$5.5234 min
48Hy4 preview8.1% ±1.3$3.072.2 h
49Muse Spark Open6.6% ±0.5$7.262.1 h
50Muse Spark 1.26.1% ±1.5$4.1478 min
50Gemini 3.5 Flash6.1% ±1.5$5.9871 min
52Beam 1.04.0% ±1.3—19 min
53Mercury 33.0%$0.42233 min
54Gemini 3.1 Pro Preview2.5% ±0.5$4.1119 min
54GPT-5.4 mini2.5% ±1.0$1.4430 min
56MiniMax M31.0% ±1.0$7.2376 min
57MiMo-V2.5-Pro0.5% ±0.5$0.2332.1 h
57Inkling0.5% ±0.5$2.5023 min
59Inkling-Small0.0%$0.32212 min
59Nemotron 3.5 Super VL 120B A12B0.0%$0.86315 min
59Mercury 2.50.0%$0.36815 min