LLM Stats (MRCR): leaderboard
Multi-Round Coreference Resolution: long-context retrieval testing coreference tracking across extended multi-turn conversations.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 93 | #152 |
| 2 | Gemini 1.5 Pro | 82.6 | #445 |
| 3 | Gemini 1.5 Flash | 71.9 | #618 |
| 4 | Gemini 2.0 Flash | 69.2 | #342 |
| 5 | Gemini 1.5 Flash-8B | 54.7 | #819 |
| 6 | MiMo-V2-Flash | 45.7 | #372 |
| 7 | Gemini 2.5 Flash | 32 | #245 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-mrcr · How It Works · Data refreshed daily, snapshot 2026-10-20.