Meta Launches Proprietary Multimodal AI Model Muse Spark
Meta has officially launched Muse Spark, its first flagship multimodal AI model developed under the newly formed Superintelligence Labs led by Alexandr Wang. Unlike the company's previous Llama series, Muse Spark is a proprietary, closed-source model designed to natively process visual and textual data with low latency on consumer hardware. The model is currently being deployed across Meta's vast ecosystem, including the Meta AI app, WhatsApp, Instagram, Facebook, and Ray-Ban Meta glasses, reaching an initial audience of three billion daily active users. This launch marks a strategic shift for Meta, moving from open-source contributions to direct competition with proprietary models from OpenAI and Anthropic. Muse Spark emphasizes agentic capabilities, enabling autonomous task execution rather than simple chat responses. The article highlights that this move establishes native multimodality as an industry baseline and shifts competitive focus toward orchestration, context management, and deployment scale. Developers are urged to reconsider their architectures as Meta integrates this intelligence directly into its platforms, potentially positioning Muse Spark as both a tool and a competitor for third-party applications.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page itself is projected from evidence records.
- Current automated evidence projection