Muse Code and Muse Spark 1.2

Muse announced the release of Muse Code in beta, a terminal coding agent powered by the new Muse Spark 1.2 model. The company said Muse Code handles complex software engineering tasks across large repositories by planning changes, writing code, and validating results. Muse Code uses async background agents that remain active throughout a session to reduce redundant information gathering and latency. A local event log records every model call, tool run, approval, and edit to make the runtime replay-exact and restart-safe after crashes. The agent ships with default skills including plan, grill, and goal for task management. Muse Spark 1.2 improves code generation, complex debugging, and codebase understanding over version 1.1. The model was co-trained with Muse Code using rejection sampled harness trajectories and recipe optimizations. Training included long-horizon tasks such as whole-repository generation and a self-improvement loop where Muse Spark 1.1 generated challenging environments for Muse Spark 1.2 to solve. Muse Spark 1.2 is available today in Muse Code and via the Meta Model API with expanded global access.