The Tool Is Meta's Answer to Anthropic's Claude Code and Openai's Codex, and Is Priced Lower Than Competitors

Meta announced Wednesday a beta version of Muse Code, a coding agent powered by its Muse Spark 1.2 AI model, designed to handle complex software engineering tasks like writing code and validating results. The tool is Meta's answer to Anthropic's Claude Code and OpenAI's Codex, and is priced lower than competitors. Meta's AI chief Alexandr Wang said in an X post that Muse Code is "one of the most affordable coding agents on the market, available globally.

The standard version of Muse Spark 1.2 costs $1.25 per million input tokens, 15 cents per million cached-input tokens, and $4.25 per million output tokens. A discounted "contributor tier requires user activity to improve Meta's products, appealing to individual developers and startups, while companies with proprietary code may prefer the standard tier. Direct comparisons are difficult since Claude Code and Codex are bundled into subscriptions: Anthropic's Pro Plan at $20 a month includes Claude Code, with Max plans at $100 to $200 a month; OpenAI's Codex is available via ChatGPT Plus at $20 a month, and its Pro plan costs $100 a month.

The launch underscores coding as a key AI battleground, as coding agents can examine software, run tests, and write code with limited supervision. During Meta's first-quarter earnings call in April, CEO Mark Zuckerberg said developer tools weren't the company's primary focus but he wasn't opposed to offering coding tools. With Muse Code, Meta aims to attract programmers who might use Claude or Codex by offering a comparable agent at a lower price.