MSU-Bench - Dialogue Reasoning Tier: leaderboard
Tier 2 of MSU-Bench (2026): six multi-speaker dialogue reasoning tasks on scene background, speaker roles, dialogue acts, question and answer structure, emotional interaction and aggregated viewpoints. MSU-Bench (2026) poses four-option questions over real multi-speaker conversations in Chinese and English, with distractors that attribute content to the wrong speaker or hallucinate it; accuracy in percent, higher is better.
Metric: Tier 2 accuracy (%), instance-level over the six multi-speaker dialogue reasoning tasks, four-option multiple choice on MSU-Bench (2,300 human-verified questions over Chinese and English multi-speaker conversations from telephone, meeting, podcast and movie audio), zero-shot with one instruction template requiring a single option letter, exact-match accuracy (x100); higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 84 | #164 |
| 2 | Gemini 2.5 Flash | 77 | #453 |
| 3 | Gemini 2.5 Pro | 74 | #266 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=msu-bench-dialogue-reasoning-tier · How It Works · Data refreshed daily, snapshot 2026-11-13.