Benchmark
MRCR v2
MRCR v2 (Multi-Round Coreference Resolution version 2) is an enhanced version of the synthetic long-context reasoning task. It extends the original MRCR…
- Modality
- text
- Categories
- long_context, reasoning, general
- Openness
- unknown
- Source
- llm_stats
- Reported scores
- 3
Reported scores
llm_stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| DiffusionGemma 26B-A4B | 0.32 | 2026-06-10 | evidence | |
| Gemini 2.5 Flash-Lite | 0.166 | 2025-06-17 | evidence | |
| Qwen3.7-Plus | Qwen | 0.917 | 2026-05-31 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.