MP-Bench - Discussion Appropriateness: leaderboard

MP-Bench (2026) evaluates voice agents mediating a multiparty dispute in its Discussion scenario. This board is appropriateness accuracy, where a turn counts only when the agent both takes the floor when required and gives a logically consistent response, judged by a GPT-5 LLM judge, in percent, higher is better.

Metric: Appropriateness accuracy (%; Discussion scenario, correct only when the agent both takes the floor when required and gives a logically consistent response, GPT-5 judge). Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked.

Top models

No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.

Interactive version: theaggregate.ai/benchmark?slug=mp-bench-discussion-appropriateness · How It Works · Data refreshed daily, snapshot 2026-11-12.