Meta enters AI coding market with launch of Muse Code

Meta launched Muse Code on Wednesday, August 5, an artificial intelligence tool designed to write and debug software autonomously. Powered by the Muse Spark 1.2 model, the tool can plan project execution, divide work into tasks, and utilize multiple sub agents to handle complex coding workflows. Meta stated that the model scored 82.9 percent on Terminal Bench 2.1, placing it near Anthropic's Claude Opus 5.

The launch marks Meta's entry into a highly competitive segment featuring tools from OpenAI, Anthropic, Google, and Microsoft. Muse Code is currently in beta and uses a pay as you go pricing model rather than a subscription, which is a strategy intended to gain market share. Meta is seeking to monetize its AI models directly to address investor concerns regarding spending.

Reports on the launch vary in their focus. One center leaning source emphasizes the financial motivations and security risks, noting that Meta is targeting the most lucrative part of the generative AI market. This source also highlights a security incident where a previous version of the model, Muse Spark, accessed a company's systems due to a configuration error. Another center leaning source focuses on the technical specifications and the broader industry shift, framing the tool as a collaborator that automates routine work to accelerate development cycles rather than a threat to developer jobs.