Wikibench

C-Eval

reasoning

C-Eval is a comprehensive Chinese evaluation suite designed to assess advanced knowledge and reasoning abilities of foundation models in a Chinese context. It comprises 13,948 multiple-choice questions across 52 diverse disciplines spanning humanities, science, and engineering, with four difficulty levels: middle school, high school, college, and professional. The benchmark includes C-Eval Hard, a subset of very challenging subjects requiring advanced reasoning abilities.

Leaderboard

Showing 18 of 18 results

  1. Qwen3.6 Plus
    93.3%
    i
  2. Qwen3.5-397B-A17B
    93.0%
    i
  3. Kimi K2 Base
    92.5%
    i
  4. Qwen3.5-122B-A10B
    91.9%
    i
  5. MiMo-V2.5-Pro
    91.5%
    i
  6. Qwen3.6-27B
    91.4%
    i
  7. Qwen3.5-27B
    90.5%
    i
  8. Qwen3.5-35B-A3B
    90.2%
    i
  9. Qwen3.6-35B-A3B
    90.0%
    i
  10. Kimi-k1.5
    88.3%
    i
  11. Qwen3.5-9B
    88.2%
    i
  12. DeepSeek-V3
    86.5%
    i
  13. Qwen3.5-4B
    85.1%
    i
  14. Qwen2 72B Instruct
    83.8%
    i
  15. Qwen2 7B Instruct
    77.2%
    i
  16. Qwen3.5-2B
    73.2%
    i
  17. Qwen3.5-0.8B
    50.5%
    i
  18. ERNIE 4.5
    40.7%
    i