τ²-bench (Telecom)
A dual-control, multi-turn tool-agent-user benchmark (telecom split) where both the AI agent and a simulated user invoke tools to coordinate and resolve technical-support troubleshooting tasks in a shared, dynamic environment.
What this benchmark measures
A dual-control, multi-turn tool-agent-user benchmark (telecom split) where both the AI agent and a simulated user invoke tools to coordinate and resolve technical-support troubleshooting tasks in a shared, dynamic environment.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is pass^1. It should be interpreted within τ²-bench (Telecom), not compared as part of a site-wide ranking.
Frequently asked
What is τ²-bench (Telecom)?
A dual-control, multi-turn tool-agent-user benchmark (telecom split) where both the AI agent and a simulated user invoke tools to coordinate and resolve technical-support troubleshooting tasks in a shared, dynamic environment. It is a tool use benchmark measured by pass^1.
What does pass^1 mean on τ²-bench (Telecom)?
τ²-bench (Telecom) reports pass^1 (%); higher is better. Scores are shown only within τ²-bench (Telecom) and are never averaged with other benchmarks.
What is the top reported τ²-bench (Telecom) score?
Claude Opus 4.6 has the top reported score on τ²-bench (Telecom): 99.3% (pass^1).
Why do τ²-bench (Telecom) scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. τ²-bench (Telecom) scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".