Meta has introduced its first AI coding agent, Muse Code, marking the company’s latest push to compete with AI heavyweights OpenAI and Anthropic in one of the fastest-growing areas of artificial intelligence.
The new tool is the latest product unveiled under Alexandr Wang, head of Meta Superintelligence Labs, who joined the company last year as part of CEO Mark Zuckerberg’s effort to reshape Meta’s AI strategy and accelerate its race against rival AI labs.
According to Wang, Muse Code is designed to do much more than autocomplete lines of code.
“You can install it with one command and then use it to take on complete software engineering tasks across a wide variety of use cases, planning changes, writing code, validating the results,” he said in an interview on Wednesday.
The assistant is capable of helping developers plan software updates, generate code and verify that changes work as intended, all from a single interface.
New tool is part of Meta’s broader push to monetize its AI spending
The launch comes as Meta continues investing billions of dollars in AI infrastructure, including data centers and computing hardware, while searching for new ways to generate returns from those investments.
Investor pressure has been mounting after Meta recently issued a weaker-than-expected revenue outlook and reported lower free cash flow for the second quarter.
Muse Code enters a market already dominated by AI coding assistants such as Anthropic’s Claude and OpenAI’s Codex, both of which are increasingly being used to automate software development tasks.
Meta says its offering allows developers to manage multiple AI agents while building applications, making it easier to handle different stages of the software development process from a single workspace.
The preview version of Muse Code works alongside the company’s newest AI model, Muse Spark 1.2.
Wang said the two were developed together, allowing the coding agent to benefit from improvements made to the underlying AI model.
Although he did not share usage figures, Wang said adoption of the Muse Spark family has been “exciting and strong.”
Rather than trying to differentiate itself solely on performance, Meta is also competing on price.
Developers can access Muse Code through a pay-as-you-go model, while a lower-cost “contributor tier” offers significantly cheaper access in exchange for allowing Meta to use submitted data to improve future versions of its AI models.
Meta introduces privacy option
For businesses concerned about privacy, Meta is also introducing a zero-data retention option, allowing developers to use the service without their data being stored for model training.
Wang said the feature is particularly important for enterprise customers.
Developers will be able to access Muse Code through Meta’s developer platform, where the company already hosts its Muse Spark API.
The latest Muse Spark model will also be available through OpenRouter, giving developers another way to integrate it into their applications.
With Muse Code, Meta is making it clear that it wants to be more than a consumer AI company. As competition intensifies, the company is expanding into developer tools in hopes of becoming a bigger player in the rapidly growing market for AI-powered software engineering.
