Wikibench

MRCR 1M

reasoning official site →

Multilingual: no

MRCR 1M is a variant of the Multi-Round Coreference Resolution benchmark designed for testing extremely long context capabilities with approximately 1 million tokens. It evaluates models' ability to maintain reasoning and attention across ultra-long conversations.

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

Showing 3 of 3 results

  1. DeepSeek-V4-Pro-Max self-reported llm-stats
    83.5%
  2. DeepSeek-V4-Flash-Max self-reported llm-stats
    78.7%
  3. Gemini 2.0 Flash-Lite self-reported llm-stats
    58.0%