Wikibench

Claw-Eval

coding

Multilingual: no

Claw-Eval tests real-world agentic task completion across complex multi-step scenarios, evaluating a model's ability to use tools, navigate environments, and complete end-to-end tasks autonomously.

Methodology

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

Leaderboard

Showing 8 of 8 results

  1. Kimi K2.6 self-reported llm-stats
    80.9%
  2. GLM-5V-Turbo self-reported llm-stats
    75.0%
  3. MiniMax M3 self-reported llm-stats
    74.5%
  4. Qwen3.7 Max self-reported llm-stats
    65.2%
  5. MiMo-V2-Pro self-reported llm-stats
    61.5%
  6. Qwen3.6-27B self-reported llm-stats
    60.6%
  7. Qwen3.6 Plus self-reported llm-stats
    58.7%
  8. MiMo-V2-Omni self-reported llm-stats
    54.8%