Muse Code is Meta‘s first programming agent and arrives as a direct move to compete with Anthropic and OpenAI. The tool, announced in New York, has been launched in beta and can already be installed on macOS and Linux.
The move confirms that Meta wants to enter the code assistant market more strongly, a segment where response speed, reasoning quality, and cost per use weigh as much as technical precision. Muse Code works alongside the Muse Spark 1.2 model, the foundation upon which the company is building its new generation of AI tools.
Muse Code and Meta’s AI plan
Meta had already been expanding its artificial intelligence offensive with recent releases such as Muse Spark 1.1, focused on deep reasoning, and Muse Image, its high-quality image generator. With Muse Code, the company takes another step toward products designed for developers and professional use.
Alexandr Wang, Meta’s Head of AI, explained to CNBC that the API maintains prices similar to those of the previous version. The pricing scheme charges $1.25 per million input tokens and $4.25 per million output tokens.
What Muse Code offers
According to available information, Muse Code functions as a terminal coding agent, meaning an assistant that can help write, review, and process programming instructions within the developer’s working environment.
- Status: beta phase.
- Compatibility: macOS and Linux.
- Associated model: Muse Spark 1.2.
- Billing model: pay-as-you-go.
Meta’s competitive bet
Meta’s entry into this niche is no coincidence. The market for programming assistants has grown rapidly with products that automate repetitive tasks, suggest code, and accelerate workflows. In this context, Muse Code seeks to position itself as an enterprise and technical alternative within the company’s ecosystem.
The launch also fits into Meta’s broader AI strategy, which includes products for mass consumption and professional tools. For developers, the deciding factor will be whether Muse Code manages to combine performance, stability, and competitive costs against already established rivals.
With this move, Muse Code becomes a key test to measure how far Meta can go in assisted programming. If the beta performs well, Muse Code could quickly gain space among technical teams; if not, Muse Code will remain just another attempt by the company to catch up with industry leaders. Even so, Muse Code marks a clear step in Meta’s offensive to contest the business of AI applied to software development.
