Meta Launches Muse Spark to Dominate AI After Llama

Meta Redefines Its AI Strategy with Muse Spark

Meta, under the leadership of Mark Zuckerberg, has decided to reposition itself in the field of artificial intelligence following the failure of its Llama models. The company has established a new team, Superintelligence Labs, by recruiting experts from competing firms to develop an AI capable of competing with the technologies of Anthropic, Google, OpenAI, and xAI. Unlike its previous projects, Meta is no longer fully relying on open source for Muse Spark.

Nine months after the launch of this ambitious project, Meta unveiled the Muse family, led by the Muse Spark model. Benchmark results shared by Alexandr Wang, the head of Superintelligence Labs, show that Muse Spark (Thinking) achieves performance comparable to that of leading competitors. In some evaluations, it even surpasses models such as Claude Opus 4.6 Max, Gemini 3.1 Pro High, GPT-5.4 Xhigh, and Grok 4.2 Reasoning.

The Artificial Analysis index, which compiles several evaluations, ranks Muse Spark in 4th place, well above the most powerful version of Llama, which is ranked 18th. This significant progress highlights Meta's advancement in the AI competition.

A Smarter and Multimodal AI Model

Muse Spark repositions Meta in the race with a faster and smarter model. It offers different modes, including the Thinking mode for complex queries. Additionally, Muse Spark can mobilize multiple AI sub-agents simultaneously and possesses multimodal perception, allowing it to understand images, such as estimating the calorie count of a dish from a photo. A shopping mode also facilitates the search for items to purchase. When a user searches for a place to visit or inquires about current events, Meta AI can generate enriched responses with information sourced from social media.

Gradual Deployment of Muse Spark

For now, Muse Spark is available only in the United States via the Meta AI app and website. In the coming weeks, it will be rolled out in other countries and on platforms such as Instagram, Facebook, Messenger, WhatsApp, as well as on Meta's smart glasses. The Ray-Ban Meta glasses will particularly benefit from the new perception capabilities of Muse Spark.

Justifying Colossal Investments

To develop Muse Spark, Meta has invested heavily, including offering bonuses to attract experts from the competition. The group has also invested $14 billion in the startup Scale AI, which specializes in data for AI, and has recruited its CEO to lead the new AI branch. These investments appear to be paying off, as the announcement of Muse Spark led to an increase in Meta's stock value. Meanwhile, Meta is investing in data centers to support the development and deployment of this new AI. These data centers are essential for effectively training and deploying Muse Spark, underscoring Meta's commitment to establishing itself as a leader in the field of artificial intelligence.