Wikibench

CyberGym

coding

Multilingual: no

CyberGym is a benchmark for evaluating AI agents on cybersecurity tasks, testing their ability to identify vulnerabilities, perform security analysis, and complete security-related challenges in a controlled environment.

Methodology

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

Leaderboard

Showing 7 of 7 results

  1. Claude Mythos Preview self-reported llm-stats
    83.1%
  2. GPT-5.5 self-reported llm-stats
    81.8%
  3. Claude Opus 4.8 self-reported llm-stats
    78.8%
  4. Claude Opus 4.6 self-reported llm-stats
    73.8%
  5. Claude Opus 4.7 self-reported llm-stats
    73.1%
  6. GLM-5.1 self-reported llm-stats
    68.7%
  7. Kimi K2.5 self-reported llm-stats
    41.3%