finance
Meta Launches Muse Spark AI Model from Superintelligence Labs
Meta announced Muse Spark, its first AI model from the Meta Superintelligence Labs, on Wednesday. The model powers the Meta AI app and website, with rollout planned for Facebook, Instagram, WhatsApp, and Messenger in coming weeks. Meta highlighted the model's efficiency in reasoning tasks, amid ongoing investments in AI infrastructure.
KO
Substrate placeholder — needs reviewAudio version
Tap play to generate a narrated version.
Meta released Muse Spark, a new artificial intelligence model developed by its Meta Superintelligence Labs. The model, originally codenamed Avocado, is the first in the company's Muse series. It became available in the standalone Meta AI app and desktop website on Wednesday.
Post by @KobeissiLetter on X
Muse Spark is designed as a small and fast model capable of reasoning through complex questions in science, math, and health. Meta stated that improved training techniques and rebuilt infrastructure enabled creation of models as capable as previous midsize variants but with significantly less compute.
The company positions it for competitive performance in multimodal perception, reasoning, health, and agentic tasks.
Labs was established following the hiring of Alexandr Wang in June, who previously served as CEO of Scale AI.
3 billion investment in Scale AI nine months ago. The labs rebuilt Meta's AI stack over the last nine months, accelerating development cycles. The release follows the disappointing performance of Meta's previous open-source models in April, which failed to attract developers.
CEO Mark Zuckerberg adjusted the strategy in response. Muse Spark marks a shift to a proprietary model, though Meta expressed intent to open-source future versions.
now powers Meta's digital assistant across its platforms.
It will integrate into Facebook, Instagram, WhatsApp, and Messenger within weeks, as well as Ray-Ban Meta AI glasses. The model will also support the Vibes AI video feature in the Meta AI app, replacing third-party models. Users can select modes based on query complexity: Instant for simple questions, Thinking for tasks like analyzing legal documents or nutritional information from photos, and Contemplating for advanced reasoning using multiple AI agents.
