tau-voice: τ-Voice (τ³-Voice)
Sierra full-duplex voice-agent benchmark on airline, retail and telecom customer-service tasks; task completion is pass@1 with each language-domain cell weighted equally.
Models ranked
2
tracked on this benchmark
Score band
68.5 – 45.7
best → lowest tracked
Snapshot trend
—
need ≥2 snapshots
Leaderboard
Tracked models ranked by % Task completion (pass@1) (higher is better).
Configuration: gemini-3.8-live-extended-thinking (Standard submission)
Notes: Official Sierra taubench.com τ³-Voice leaderboard (Overall tab, Standard, All languages): '#5 gemini-3.8-live-extended-thinking Google 68.5% 15.8% 37.8%' (task completion / interaction / generation quality). Headline is a multilingual average: task completion is pass@1 over the frame, each language–domain cell weighted equally. Google's launch blog states 68.6%. Retrieved 2026-11-02.
Configuration: GPT-Realtime-2.1 (reasoning effort not stated)
Notes: OpenAI staff announcement on the official OpenAI Developer Community (post by sps, 2026-09-10 10:57 PT): 'GPT-Live-1 completed 83.6% of Tau3 tasks on the first attempt, versus 45.7% for GPT-Realtime-2.1.' OpenAI staff post; settings not stated. Retrieved 2026-11-02.
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
- The model list covers the same version family you can actually deploy today.
- Top candidates overlap with your required routing and feature requirements.
Be cautious when
- There is only one benchmark snapshot or the dataset appears stale.
- The benchmark metric direction is opposite of your decision objective.
- The score difference between options is narrow and likely within implementation variance.
Related benchmarks
Last reviewed: Nov 2, 2026