Which AI model scores highest on τ²-bench?

Gemini 3.1 Pro by Google holds the best τ²-bench (telecom) result among tracked models, at 99.3% (released Feb 19 2026). Higher scores are better on this benchmark.

What are the top 5 models on τ²-bench?

1. Gemini 3.1 Pro (Google) — 99.3%; 2. Gemini 3.0 Pro (Google) — 98%; 3. GPT-5 (OpenAI) — 96.7%; 4. GPT-5 mini (OpenAI) — 74.1%; 5. GPT-5 nano (OpenAI) — 35.5%.

How many models have a published τ²-bench score?

5 tracked models have a published τ²-bench (telecom) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Agentic tasks

τ²-benchtelecom

task success in the telecom domain of the second-generation customer-service evaluation

retail airline telecom telecom, adjusted prompt

Rankings

Higher is better

τ²-bench — frequently asked questions

: task success in the telecom domain of the second-generation customer-service evaluation
: Gemini 3.1 Pro by Google holds the best τ²-bench (telecom) result among tracked models, at 99.3% (released Feb 19 2026). Higher scores are better on this benchmark.
: 1. Gemini 3.1 Pro (Google) — 99.3%; 2. Gemini 3.0 Pro (Google) — 98%; 3. GPT-5 (OpenAI) — 96.7%; 4. GPT-5 mini (OpenAI) — 74.1%; 5. GPT-5 nano (OpenAI) — 35.5%.
: 5 tracked models have a published τ²-bench (telecom) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.