Last released Sep 6, 2026
Benchmarking suite for evaluating agentic and roleplay capabilities of small VLLMs