Psi-Bench - Viewpoint Debate: leaderboard
Viewpoint debate in Psi-Bench (2026): how far the model shifts the opinion of a simulated, persona-profiled client in Change My View style debates, rated 1 to 9 by a DeepSeek-V3.2 judge that also plays the client. Higher is better.
Metric: Persuasion effect on the client's opinion, Change My View debates (1-9; DeepSeek-v3.2 judge, which also plays the persona-profiled client; three rounds after the client's opening message, client profile hidden from the model). Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 | 6.12 | #229 |
| 2 | Gemini 3.1 Pro (Preview) | 5.89 | #94 |
| 3 | DeepSeek V4 Pro | 5.71 | #188 |
| 4 | GPT-5 Mini | 5.27 | #348 |
| 5 | DeepSeek V3.2 | 4.81 | #406 |
| 6 | Gemini 3 Flash | 4.68 | #164 |
| 7 | Grok 4 Fast | 4.57 | #441 |
| 8 | Qwen 3 32B | 4.3 | #960 |
| 9 | Qwen 3 Next 80B A3B | 4.2 | #757 |
| 10 | Qwen 3 8B | 3.51 | #1830 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=psi-bench-viewpoint-debate · How It Works · Data refreshed daily, snapshot 2026-11-13.