Hy-MultiTurn - Strict Accuracy: leaderboard
The strict accuracy of Hy-MultiTurn (Tencent Hunyuan, 2026): the percent of its 209 long Chinese multi-turn dialogue tasks whose final response satisfies every checklist item, averaged over three rolls and three LLM judges. Higher is better.
Metric: Strict accuracy (%): share of tasks whose every checklist item is satisfied, 209 long Chinese multi-turn dialogues (12 to 76 turns), three rolls per task averaged, three LLM judges (GPT-5.2, Qwen 3.6, DeepSeek-V4 Pro) averaged, a deterministic checker for precise execution; higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (High) | 41.1 | #49 (GPT-5.5) |
| 2 | Grok 4.5 (High) | 39.9 | #82 (Grok 4.5) |
| 3 | Kimi K3 (Max) | 35.9 | #54 (Kimi K3) |
| 4 | GPT-5.4 (High) | 35.5 | #123 (GPT-5.4) |
| 5 | Claude Opus 4.6 (High) | 32.7 | #102 (Claude Opus 4.6) |
| 6 | Gemini 3.1 Pro (Preview) (High) | 32.4 | #94 (Gemini 3.1 Pro (Preview)) |
| 7 | Claude Opus 4.7 (High) | 30.4 | #79 (Claude Opus 4.7) |
| 8 | Kimi K2.6 | 25.9 | #178 |
| 9 | Doubao-Seed-2.1-Pro (High) | 25.5 | |
| 10 | Hy3 (High) | 24.4 | #195 (Hy3) |
| 11 | Kimi K2.5 (Thinking) | 21 | #256 (Kimi K2.5) |
| 12 | Kimi K2.6 (Non-reasoning) | 20 | #178 (Kimi K2.6) |
| 13 | Qwen 3.6 Plus (Thinking) | 19.5 | #222 (Qwen 3.6 Plus) |
| 14 | Kimi K2.5 (Non-reasoning) | 17.9 | #256 (Kimi K2.5) |
| 15 | Hy3-preview (High) | 13.8 | #316 (Hy3-preview) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=hy-multiturn-strict-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-13.