Meta enters the AI coding wars with Muse Spark 1.2 and Muse Code with persistent async background agents

Meta today released Muse Code, a terminal-based AI coding agent now in beta, alongside Muse Spark 1.2, a coding-focused update to its Muse Spark family of frontier models. Meta co-founder and CEO Mark Zuckerberg announced the launch in a post on X. The release puts the company in direct competition with Anthropic's Claude Code, OpenAI's Codex, and other agentic coding tools. Muse Code features persistent async background agents that remain active throughout a session rather than spawning fresh for each task. Meta said this design avoids redundant information gathering and reduces latency. The agent can fan out to parallel sub-agents working in isolated git worktrees for large jobs. Every model call and edit is appended to a local event log before execution, making the runtime replay-exact and restart-safe. The tool is installable on macOS or Linux via a single curl command but requires a Meta account and billing details. Muse Spark 1.2 was co-trained with Muse Code using rejection-sampled harness trajectories and a self-improvement loop. Meta published benchmark charts showing the model scoring 82.9% on Terminal-Bench 2.1, 59.3% on DeepSWE 1.1, and 70.6% on an internal coding benchmark. In all three evaluations, Anthropic's Opus 5 led the results. Muse Spark 1.2 improved on its predecessor by 6.7 points on Terminal-Bench and 6.3 points on DeepSWE.