An open benchmark for evaluating task execution and communication by language-model agents in dual-control human-agent environments.
Browse the latest scores, model modes, release dates, and parameter sizes for τ²-Bench.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 Claude Opus 4.6 Extended ThinkingTools | 91.89 | 2026-02-05 | Unknown | Closed | |
 Gemini 3.1 Pro Preview Thinking Level · HighTools | 90.80 | 2026-02-20 | Unknown | Closed | |
Gemini 3.0 Flash Thinking EnabledTools | 90.20 | 2025-12-17 | Unknown | Closed | |
4 |  GLM-5 Thinking EnabledTools | 89.70 | 2026-02-11 | 744B | Free Commercial |
5 |  Step 3.5 Flash Thinking EnabledTools | 88.20 | 2026-02-02 | 196B | Free Commercial |
6 | GLM-4.7 Thinking EnabledTools | 87.40 | 2025-12-22 | 358B | Free Commercial |
7 |  Qwen3.5-397B-A17B Thinking EnabledTools | 86.70 | 2026-02-16 | 39.7B | Free Commercial |
8 | Gemini 3.0 Pro (Preview 11-2025) Thinking EnabledTools | 85.40 | 2025-11-18 | Unknown | Closed |
9 | Claude Sonnet 4.5 Thinking EnabledTools | 84.70 | 2025-09-30 | Unknown | Closed |
10 |  Grok 4.1 Fast Thinking EnabledTools | 82.71 | 2025-11-19 | Unknown | Closed |
11 | Qwen3-Max-Thinking Thinking EnabledTools | 82.10 | 2026-01-26 | 1000B | Closed |
12 |  GPT-5.2 Thinking Level · Extra HighTools | 82.00 | 2025-12-11 | Unknown | Closed |
13 | Opus 4.5 Extended ThinkingTools | 81.99 | 2025-11-25 | Unknown | Closed |
14 |  DeepSeek V3.2 Thinking EnabledTools | 80.30 | 2025-12-01 | 671B | Free Commercial |
15 | GPT-5 Thinking EnabledTools | 80.00 | 2025-08-07 | Unknown | Closed |
16 | GLM-4.7-Flash Thinking EnabledTools | 79.50 | 2026-01-19 | 31B | Free Commercial |
17 | Qwen3.5-27B Thinking EnabledTools | 79.00 | 2026-02-25 | 27B | Free Commercial |
18 |  ERNIE 5.0 Thinking Enabled | 78.79 | 2025-11-13 | 2400B | Closed |
19 |  MiniMax M2 Thinking EnabledTools | 77.20 | 2025-10-27 | 230B | Free Commercial |
20 |  Gemma 4 31B Thinking EnabledTools | 76.90 | 2026-04-02 | 30.7B | Free Commercial |
21 | GLM-4.6 Thinking EnabledTools | 75.90 | 2025-09-30 | 355B | Free Commercial |
22 | Qwen3 Max (Preview) Standard ModeTools | 74.00 | 2025-09-05 | Unknown | Closed |
23 | Claude Opus 4 Thinking EnabledTools | 72.50 | 2025-05-23 | Unknown | Closed |
24 | Qwen3 Max (Preview) Thinking EnabledTools | 72.00 | 2025-09-05 | Unknown | Closed |
25 | Claude Sonnet 4.5 Standard ModeTools | 71.00 | 2025-09-30 | Unknown | Closed |
26 | Gemma 4 26B A4B Thinking EnabledTools | 68.20 | 2026-04-02 | 25.2B | Free Commercial |
27 | DeepSeek V3.2-Exp Thinking EnabledTools | 66.70 | 2025-09-29 | 671B | Free Commercial |
28 |  Kimi K2 Standard ModeTools | 64.30 | 2025-07-11 | 1000B | Free Commercial |
29 | Kimi K2 Thinking EnabledTools | 64.30 | 2025-07-11 | 1000B | Free Commercial |
30 | Claude Sonnet 3.7 Thinking EnabledTools | 61.80 | 2025-02-25 | Unknown | Closed |
