agentscope-ai/pawbench

PawBench: A comprehensive agent benchmark with 150 tasks covering coding, data analysis, tool use, reasoning, safety, and multimodal capabilities.

Published 6/11/2026 by

harbor run -d agentscope-ai/pawbench

Displaying 0 of 150 tasks