MOSAIC: A Unified Platform for Cross-Paradigm Comparison and Evaluation of Homogeneous and Heterogeneous Multi-Agent RL, LLM, VLM, and Human Decision-Makers

MOSAIC is a new open-source platform that enables cross-paradigm comparison and evaluation of different AI agents including reinforcement learning, large language models, vision-language models, and human decision-makers within the same environment. The platform introduces three key technical contributions: an IPC-based worker protocol, operator abstraction for unified interfaces, and a deterministic evaluation framework for reproducible research.

• →MOSAIC bridges the gap between isolated AI paradigms by allowing RL, LLM, VLM, and human agents to operate in the same environment.
• →The platform uses an IPC-based worker protocol that wraps different frameworks as isolated subprocess workers while maintaining their native logic.
• →It provides a unified operator abstraction that creates consistent interfaces regardless of the underlying agent type.
• →The evaluation framework offers both manual lock-step inspection and automated scripted evaluation modes for reproducible experiments.
• →MOSAIC is released as an open-source, visual-first platform to facilitate cross-paradigm research across multiple AI communities.