Wikibench

IFBench

general

Multilingual: no

Instruction Following Benchmark evaluating model's ability to follow complex instructions

Methodology

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

Leaderboard

Showing 19 of 19 results

  1. Hermes 3 70B self-reported llm-stats
    81.2%
  2. Nova 2 Pro self-reported llm-stats
    80.2%
  3. Qwen3.7 Max self-reported llm-stats
    79.1%
  4. Qwen3.5-27B self-reported llm-stats
    76.5%
  5. Qwen3.5-397B-A17B self-reported llm-stats
    76.5%
  6. Qwen3.5-122B-A10B self-reported llm-stats
    76.1%
  7. MAI-Code-1-Flash self-reported llm-stats
    75.0%
  8. Qwen3.6 Plus self-reported llm-stats
    74.2%
  9. Nemotron 3 Super (120B A12B) self-reported llm-stats
    72.6%
  10. Mercury 2 self-reported llm-stats
    71.0%
  11. Nova 2 Lite self-reported llm-stats
    70.8%
  12. Qwen3.5-35B-A3B self-reported llm-stats
    70.2%
  13. MiniMax M2.1 self-reported llm-stats
    70.0%
  14. GPT OSS 120B High self-reported llm-stats
    69.5%
  15. MAI-Thinking-1 self-reported llm-stats
    69.0%
  16. Mistral Medium 3.5 self-reported llm-stats
    69.0%
  17. Nova 2 Omni self-reported llm-stats
    68.7%
  18. K-EXAONE-236B-A23B self-reported llm-stats
    67.3%
  19. Nova 2 Sonic self-reported llm-stats
    37.5%