New: connect Claude & other AIs to GenAIList over MCP — research the catalog and contribute to the shared knowledge base. Learn how →
τ²-Bench (TAU2) benchmark
AI model leaderboard for the τ²-Bench (TAU2) benchmark. Compare how large language models score on τ²-Bench (TAU2), see the full ranking, and understand what this AI benchmark measures. Gemini 3 Pro currently leads with 85.4. Two-agent dual-control variant of τ-Bench — dialogue benchmark where both user and assistant are role-played, increasing realism.
Leaderboard
| # | Model | Organization | Score | Variant | Source |
|---|---|---|---|---|---|
| #1 | Gemini 3 Pro | Google DeepMind | 85.4 | cited-deepseek-v3-2 | official ↗ |
| #2 | Claude Sonnet 4.5 | Anthropic | 84.7 | cited-deepseek-v3-2 | official ↗ |
| #3 | DeepSeek-V3.2 | DeepSeek | 80.3 | thinking | official ↗ |
| #4 | GPT-5 | OpenAI | 80.2 | cited-deepseek-v3-2 | official ↗ |
| #5 | MiniMax-M2 | MiniMax | 76.9 | cited-deepseek-v3-2 | official ↗ |
| #6 | Kimi K2 Thinking | Moonshot | 74.3 | thinking-cited-deepseek-v3-2 | official ↗ |
Frequently asked questions about τ²-Bench (TAU2)
What is the τ²-Bench (TAU2) benchmark?
Two-agent dual-control variant of τ-Bench — dialogue benchmark where both user and assistant are role-played, increasing realism.
How is the τ²-Bench (TAU2) benchmark scored?
τ²-Bench (TAU2) is scored using the accuracy metric, where a higher score is better. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.
Which AI model scores highest on τ²-Bench (TAU2)?
As of the latest reported scores on GenAIList, Gemini 3 Pro achieves the highest result on τ²-Bench (TAU2) with a score of 85.4.
Is a higher τ²-Bench (TAU2) score better?
Yes. On τ²-Bench (TAU2) a higher score indicates better performance, so models near the top of the leaderboard are the strongest.