Vals AI Terminal-Bench 4.0: leaderboard
Terminal-Bench 4.0 from Vals AI: long-horizon terminal work across seven domains (software engineering, science, machine learning, operations, hardware design, security and media production), scored pass@1 on the share of tasks completed.
Metric: Accuracy (%). Source: www.vals.ai. Saturation forecast: Around May 2027. 60 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 (Max) | 65.15 | #3 (Claude Opus 5.5) |
| 2 | Claude Sonnet 5.5 (Max) | 64.14 | #14 (Claude Sonnet 5.5) |
| 3 | Kimi K3.1 Pro (Max) | 63.64 | #17 (Kimi K3.1 Pro) |
| 4 | Gemini 4 Carbon (High) | 63.13 | #1 (Gemini 4 Carbon) |
| 5 | GPT-6 Astra (Max) | 59.6 | #6 (GPT-6 Astra) |
| 6 | Claude Fable 5.1 (Max) | 58.08 | #10 (Claude Fable 5.1) |
| 7 | Muse Wonder Full (Max) | 58.08 | #12 (Muse Wonder Full) |
| 8 | Gemini 4 Argon (High) | 57.58 | #9 (Gemini 4 Argon) |
| 9 | GPT-6.1 Sol (Max) | 55.05 | #7 (GPT-6.1 Sol) |
| 10 | Claude Opus 5 (Max) | 53.54 | #16 (Claude Opus 5) |
| 11 | Muse Wonder (Max) | 49.5 | #18 (Muse Wonder) |
| 12 | GPT-6.1 Terra (Max) | 49.5 | #15 (GPT-6.1 Terra) |
| 13 | GPT-6 Sol (Max) | 44.44 | #19 (GPT-6 Sol) |
| 14 | Claude Fable 5 (Max) | 41.41 | #20 (Claude Fable 5) |
| 15 | hy4 | 39.39 | #13 |
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-terminal-bench-4-0 · How It Works · Data refreshed daily, snapshot 2026-11-05.