Wikibench

OSWorld-Verified

multimodal official site →

Multilingual: no

OSWorld-Verified is a verified subset of OSWorld, a scalable real computer environment for multimodal agents supporting task setup, execution-based evaluation, and interactive learning across Ubuntu, Windows, and macOS.

Methodology

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

Leaderboard

Showing 15 of 15 results

  1. Claude Opus 4.8 self-reported llm-stats
    83.4%
  2. Claude Mythos Preview self-reported llm-stats
    79.6%
  3. GPT-5.5 self-reported llm-stats
    78.7%
  4. Gemini 3.5 Flash self-reported llm-stats
    78.4%
  5. Claude Opus 4.7 self-reported llm-stats
    78.0%
  6. GPT-5.4 self-reported llm-stats
    75.0%
  7. Kimi K2.6 self-reported llm-stats
    73.1%
  8. GPT-5.4 mini self-reported llm-stats
    72.1%
  9. MiniMax M3 self-reported llm-stats
    70.1%
  10. GPT-5.3 Codex self-reported llm-stats
    64.7%
  11. Qwen3.6 Plus self-reported llm-stats
    62.5%
  12. Qwen3.5-122B-A10B self-reported llm-stats
    58.0%
  13. Qwen3.5-27B self-reported llm-stats
    56.2%
  14. Qwen3.5-35B-A3B self-reported llm-stats
    54.5%
  15. GPT-5.4 nano self-reported llm-stats
    39.0%