Meta's Muse Spark 1.1: New AI Model for Developers

Meta has unveiled Muse Spark 1.1, its latest in-house artificial intelligence model, expanding public access to the technology. This move steps up Meta's efforts to compete with rivals in enterprise AI and agentic applications. The updated model follows the original Muse Spark, which had limited access. Now, developers can join a waitlist and integrate the model through a public developer preview. Meta's Chief AI Officer, Alexandr Wang, stated Muse Spark 1.1 is the company's "strongest model for agentic and coding work yet." API access will initially remain within Meta's own ecosystem. The company is also positioning this model as a lower-cost alternative. New API users will receive $20 in free credits, with pricing at $1.25 per million input tokens and $4.25 per million output tokens. The model is trained for software development tasks and AI agents. This launch follows Muse Image, Meta's AI image generation model, reflecting a broader strategy to expand proprietary AI products. Meta is increasing AI infrastructure investments and, while committed to open-source AI, is now offering proprietary models through commercial APIs. An open-source version of Muse Spark is in development, and Meta is also training a more advanced model, code-named Watermelon. This development signals Meta's push to become a major player in the competitive AI market.