EYT-Bench (PersonaMem-v2) - Latent Intent: leaderboard
EYT-Bench (Enjoy Your Talk, 2026) evaluates LLMs as multi-turn conversational partners with a persona-grounded user simulator and an independent LLM judge; each turn the target model predicts the user state before replying. Per-turn latent-intent label accuracy in percent on the PersonaMem-v2 pool of free-text personas distilled from real conversations; higher is better.
Metric: Per-turn latent-intent accuracy (%; the target model first predicts the user latent-intent label from 8 classes, scored against the simulator gold label, before generating its reply; multi-turn dialogues of up to 10 turns with a persona-grounded Gemini 3.1 Pro user simulator and a Gemini 3.1 Pro judge (reasoning effort high, temperature 0); PersonaMem-v2 pool (100 dialogues; personas are free-text paragraphs distilled from long real user-assistant interactions)). Source: arxiv.org. Saturation forecast: Around February 2027. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro | 79.9 | #188 |
| 2 | DeepSeek V4 Flash | 70.1 | #258 |
| 3 | GPT-5.5 | 47.9 | #49 |
| 4 | Gemini 3.1 Pro (Preview) (Thinking) | 47.3 | #94 (Gemini 3.1 Pro (Preview)) |
| 5 | Claude Opus 4.7 | 41.5 | #79 |
| 6 | Claude Sonnet 4.6 | 24.9 | #161 |
| 7 | Qwen 3.5 397B A17B | 15.3 | #246 |
| 8 | Qwen 3.5 35B A3B | 14.2 | #482 |
| 9 | Qwen 3.5 27B | 13.1 | #359 |
| 10 | Seed 2.0 Pro | 12 | #211 |
| 11 | Seed 2.0 Mini | 11.7 | #508 |
| 12 | Seed 2.0 Lite | 7.7 | #289 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=eyt-bench-personamem-v2-latent-intent · How It Works · Data refreshed daily, snapshot 2026-11-13.