QwenWorldBench

reasoning

QwenWorldBench is Qwen's internal benchmark for evaluating LLMs as world models that simulate agentic environments across Terminal, SWE, MCP, Search, OS, Android, and Web domains.