MM-ClawBench

coding

MM-ClawBench evaluates models on MiniMax's Claw-style agent benchmark, measuring practical agentic task completion quality in real-world OpenClaw usage scenarios.