MRCR 128K (8-needle)

reasoning official site →

MRCR (Multi-Round Coreference Resolution) at 128K context length with 8 needles. Models must navigate long conversations to reproduce specific model outputs, testing attention and reasoning across 128K-token contexts with 8 items to retrieve.

Methodology

Imported from llm-stats public benchmark metadata. Modality: text. Max score: 1. Categories: general, long_context, reasoning. Language: en. Verified by llm-stats: no.

Leaderboard

  1. Qwen3.7 Max self-reported llm-stats
    90.4%
  2. MiniCPM-SALA self-reported llm-stats
    10.1%