JaxMARL is a multi-agent reinforcement learning library that combines high-performance JAX computation with standardized MARL environments like Overcooked, Hanabi, and MPE. It provides pre-built algorithms (QMIX, MAPPO, QMIX, etc.) and benchmarking capabilities for multi-agent RL research. The library includes a walkthrough notebook and academic citations, targeting researchers working on coordination, competition, and communication in multi-agent systems.