Wikibench

MCP Atlas

coding

Multilingual: no

MCP Atlas is a benchmark for evaluating AI models on scaled tool use capabilities, measuring how well models can coordinate and utilize multiple tools across complex multi-step tasks.

Methodology

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

Leaderboard

Showing 20 of 21 results

  1. Gemini 3.5 Flash self-reported llm-stats
    83.6%
  2. Claude Opus 4.8 self-reported llm-stats
    82.2%
  3. Claude Opus 4.7 self-reported llm-stats
    77.3%
  4. Qwen3.7 Max self-reported llm-stats
    76.4%
  5. GPT-5.5 self-reported llm-stats
    75.3%
  6. MiniMax M3 self-reported llm-stats
    74.2%
  7. Qwen3.6 Plus self-reported llm-stats
    74.1%
  8. DeepSeek-V4-Pro-Max self-reported llm-stats
    73.6%
  9. GLM-5.1 self-reported llm-stats
    71.8%
  10. Gemini 3.1 Pro self-reported llm-stats
    69.2%
  11. DeepSeek-V4-Flash-Max self-reported llm-stats
    69.0%
  12. GLM-5 self-reported llm-stats
    67.8%
  13. GPT-5.4 self-reported llm-stats
    67.2%
  14. Claude Opus 4.6 self-reported llm-stats
    62.7%
  15. Claude Opus 4.5 self-reported llm-stats
    62.3%
  16. Claude Sonnet 4.6 self-reported llm-stats
    61.3%
  17. GPT-5.2 self-reported llm-stats
    60.6%
  18. GPT-5.4 mini self-reported llm-stats
    57.7%
  19. Gemini 3 Flash self-reported llm-stats
    57.4%
  20. GPT-5.4 nano self-reported llm-stats
    56.1%