τ²-bench (Telecom)
Agents · 2025-06-09
The telecom domain of τ²-bench evaluates collaborative troubleshooting in its dual-control setup. Scores report end-to-end task success.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 99.3 |
| GLM-5.2 | 99.1 |
| JT-35B-Flash | 99.1 |
| GPT-5.4 | 98.9 |
| GLM 4.7 Flash | 98.8 |
| GPT-5.2 | 98.7 |
| Claude Fable 5 | 98.5 |
| GLM 5 Turbo | 98.5 |
| GLM 5V Turbo | 98.5 |
| Step 3.7 Flash | 98.5 |
| GLM-5 | 98.2 |
| Gemini 3 Pro Preview | 98.0 |
| GPT-5.5 | 98.0 |
| Sonnet 4.6 | 97.9 |
| GLM-5.1 | 97.7 |
| Grok 4.3 | 97.7 |
| Qwen3.6 Plus (2026-04-02) | 97.7 |
| Grok 4.20 | 96.5 |
| DeepSeek-V4-Pro | 96.2 |
| GLM-4.7 | 95.9 |
| Kimi K2.5 | 95.9 |
| Kimi K2.6 | 95.9 |
| Qwen3.6-Max-Preview | 95.9 |
| DeepSeek-V4-Flash | 95.6 |
| Gemini 3.5 Flash | 95.6 |
| Qwen3.5 397B A17B | 95.6 |
| MiniMax M2.5 | 95.3 |
| Qwen3.6 35B A3B | 95.3 |
| MiMo-V2-Flash | 95.0 |
| MiMo-V2-Pro | 95.0 |
| Qwen3.7-Max | 94.7 |
| Opus 4.8 | 94.4 |
| Step 3.5 Flash | 94.4 |
| MiMo-V2.5-Pro | 94.2 |
| Mistral Medium 3.5 | 94.2 |
| Qwen3.6 27B | 94.2 |
| Qwen3.5-27B | 93.9 |
| Qwen3.5 122B A10B | 93.6 |
| GPT-5.4 Mini | 93.4 |
| Grok 4.1 Fast | 93.3 |