TAU3-Bench
Can the model complete substantial programming work under this benchmark's agent setup?
Coding5 ranked models5 reported values1 reportsHigher is better
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | Qwen3.5-35B-A3BQwen | 68.9% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
| 2 | Qwen3.5-27BQwen | 68.4% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
| 3 | Gemma4-31BGoogle | 67.5% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
| 4 | Qwen3.6-35B-A3BQwen | 67.2% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
| 5 | Gemma4-26B-A4BGoogle | 59.0% | Reported configurationQwen report1 value · 1 reportApr 15, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
No cost-linked effort sweep for this version.
Definition and comparison boundarypublic methodology
Can the model complete substantial programming work under this benchmark's agent setup?
Software implementation, debugging, or repository work under the published evaluation protocol. Higher is better. The value is the percentage reported in this lab's table.
Harness, tools, prompts, attempt count, and benchmark version can materially change the result. Compare within one reporting context.
Official user model GPT-5.2 at low reasoning effort plus default BM25 retrieval.
Method / source