Meta Joins AI Coding Competition with Innovative ‘Muse Code’ Agent
Meta Platforms has officially entered the AI coding arena with the launch of Muse Code, its pioneering AI coding agent. Released in beta on August 5, 2026, Muse Code presents a direct challenge to Anthropic’s Claude Code and OpenAI’s Codex. CEO Mark Zuckerberg unveiled the release on X, portraying it as a terminal-based agent proficient in executing comprehensive software engineering tasks across vast code repositories.
Empowered by the novel Muse Spark 1.2 model — the latest addition to Meta’s coding-centric AI suite — Muse Code is capable of planning changes, writing code, validating results, and even orchestrating multiple persistent background sub-agents. This allows it to address complex problems in parallel, making it ideal for long-horizon software engineering tasks.
Meta has priced Muse Code competitively to capture market share. The pay-as-you-go tier is priced at $1.25 per million input tokens and $4.25 per million output tokens. An ultra-affordable contributor tier is available at just $0.30 per million tokens, provided users allow Meta to use their submitted code for model training. Internally, the tool already boasts 7,000 active Meta engineers who use it weekly, generating over 800 fixes that have enhanced model performance.
The launch of Muse Code comes at a time when Meta is facing investor pressure to capitalize on its substantial AI investments, following a recent stock decline and dwindling free cash flow reported in Q2 earnings. Muse Code is now available for immediate developer testing via a straightforward command-line installation.
Source: CNBC – Meta unveils first AI coding agent to challenge Anthropic and OpenAI