Agentic benchmark
Terminal-Bench 4.0 leaderboard
Every model the catalog carries a published Terminal-Bench 4.0 value for, ranked by that value.
The current Terminal-Bench release measures difficult computer work after recalibrating task resources, fixing unstable tasks, and removing tasks that no longer separate frontier systems.
Terminal-Bench 4.0 ranking
7 models with a published Terminal-Bench 4.0 value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Task completion rate across 5 trials per task |
|---|---|---|---|
| 1 |  Claude Mythos 5.1 | Anthropic | 60.90 |
| 2 |  GPT-6 Astra | OpenAI | 57.90 |
| 3 | Claude Fable 5.1 | Anthropic | 55.80 |
| 4 | UNPareto 26.9 | Unbiased | 51.00 |
| 5 |  DeepSeek V4.1 Flash | DeepSeek | 31.20 |
| 6 |  SWE-2 | Cognition | 27.30 |
| 7 |  Gemini 3.8 Flash | 19.10 |
Evidence key: Observed
Rows are ordered by the value Terminal-Bench 4.0 published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.