Meta launches Muse Code to challenge OpenAI and Anthropic

Meta launched Muse Code, a coding tool powered by Muse Spark 1.2, and put it directly into the fight for developer loyalty that is shaping the AI market. The new agent is designed to help developers write and debug software, making it another bid by Meta to turn its model work into something coders use inside daily workflows rather than a name buried in a product list.

The release ties Meta’s coding push to Muse Spark 1.2, which now appears alongside Muse Code on the company’s developer AI pages and sits behind the Meta Model API. Meta had already started widening the Muse line on July 9, when it debuted Muse Spark 1.1 with a preview open to developers. The company has described Muse Spark as the first model in the Muse family from Meta Superintelligence Labs, and it says the system supports tool use, visual chain of thought and multi-agent orchestration, features aimed at more complex software work.

That matters because coding assistants have become one of the clearest places where generative AI can show immediate business value. If Muse Code saves time on repetitive tasks, explains logic cleanly and speeds debugging, it could help Meta win a spot in developer habits and enterprise workflows. If it is slow, brittle or difficult to integrate, coders will compare it quickly with offerings from OpenAI, Anthropic, Microsoft and Google, where the battle is increasingly about who can become the default tool inside the editor.

For enterprises, the stakes go beyond convenience. A coding agent that plugs into existing systems can lower the barrier to building software, but it also raises questions about privacy, reliability and control over code generation. Meta’s safety and preparedness report says Muse Spark is covered by the company’s Advanced AI Scaling Framework and was evaluated for catastrophic-risk domains including chemical and biological risk, cybersecurity and loss of control. That kind of scrutiny is becoming part of the sales pitch as AI tools move closer to production environments.

Meta’s timing also shows how aggressively the company is trying to claim a larger role in AI infrastructure, not just social media and advertising. By attaching Muse Code to Muse Spark 1.2, Meta is signaling that its model stack is meant to evolve into practical developer products. The real competition now is not simply who can launch another model, but whose assistant developers trust enough to build around every day.