Breaking News
Meta has introduced Muse Spark, the first model in a new series of large language models developed by its Meta Superintelligence Labs, as part of its push toward building more advanced, personalized AI assistants.
The new model is designed to be smaller and faster while still capable of handling complex reasoning tasks across domains such as science, mathematics, and health. Muse Spark now powers the upgraded Meta AI assistant available through the Meta AI app and web platform.
Meta said the model is part of a broader strategy to develop “personal superintelligence,” aiming to create AI systems that can assist users in everyday tasks with greater autonomy and contextual understanding.
A key feature of the updated assistant is its ability to deploy multiple AI “subagents” simultaneously to handle different aspects of a task. For example, when planning a trip, separate agents can work in parallel on itinerary creation, destination comparison, and activity recommendations to deliver faster and more comprehensive results.
Muse Spark also introduces enhanced multimodal capabilities, allowing the assistant to interpret images alongside text. Users can, for instance, scan products, analyze food items, or compare options visually, with the AI providing contextual insights without requiring detailed written input.
The model has also been trained to better handle health-related queries, with input from medical professionals to improve the quality and clarity of responses to common health questions.
In addition, Meta AI now includes features such as visual coding—enabling users to generate simple applications or games from prompts—and a shopping mode that integrates recommendations from creators and communities across Meta’s platforms.
The launch reflects intensifying competition in the AI space, as major technology companies race to build more capable, multimodal, and agent-driven systems that move beyond chatbots toward more proactive digital assistants.
Meta said Muse Spark is the first in a series of models, with more advanced versions already in development as it continues to scale its AI capabilities.
The new model is designed to be smaller and faster while still capable of handling complex reasoning tasks across domains such as science, mathematics, and health. Muse Spark now powers the upgraded Meta AI assistant available through the Meta AI app and web platform.
Meta said the model is part of a broader strategy to develop “personal superintelligence,” aiming to create AI systems that can assist users in everyday tasks with greater autonomy and contextual understanding.
A key feature of the updated assistant is its ability to deploy multiple AI “subagents” simultaneously to handle different aspects of a task. For example, when planning a trip, separate agents can work in parallel on itinerary creation, destination comparison, and activity recommendations to deliver faster and more comprehensive results.
Muse Spark also introduces enhanced multimodal capabilities, allowing the assistant to interpret images alongside text. Users can, for instance, scan products, analyze food items, or compare options visually, with the AI providing contextual insights without requiring detailed written input.
The model has also been trained to better handle health-related queries, with input from medical professionals to improve the quality and clarity of responses to common health questions.
In addition, Meta AI now includes features such as visual coding—enabling users to generate simple applications or games from prompts—and a shopping mode that integrates recommendations from creators and communities across Meta’s platforms.
The launch reflects intensifying competition in the AI space, as major technology companies race to build more capable, multimodal, and agent-driven systems that move beyond chatbots toward more proactive digital assistants.
Meta said Muse Spark is the first in a series of models, with more advanced versions already in development as it continues to scale its AI capabilities.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
