Meta debuts new AI model, attempting to catch up to Google, OpenAI

Meta Facebook parent OpenAI is launching its first major AI model since the costly hire of Scale AI’s Alexandr Wang nine months ago as it aims to gain a foothold in a market dominated by Anthropic and Anthropic. Google.

The artificial intelligence model announced on Wednesday, called Muse Spark and originally codenamed Avocado, is the first model of the company’s new Muse series, developed by Meta Superintelligence Labs, the artificial intelligence unit headed by Wang. Wang joined Meta in June as part of the company’s $14.3 billion investment in Scale AI, where he is CEO.

Meta is desperate to regain momentum in the competitive AI market after the disappointing launch of its latest open-source models last April. The release failed to attract developer attention, leading CEO Mark Zuckerberg to change his strategy.

“Over the past nine months, Meta Super Intelligence Labs has rebuilt our AI stack from the ground up, moving faster than any development cycle we have run before,” said Meta. blog post on Wednesday. “This first model is small and fast by design, yet capable enough to answer complex questions in science, mathematics and healthcare. It is a strong foundation and the next generation is already in development.”

Meta isn’t positioning the Muse Spark as a premium model; instead, it highlights her efficiency and “competitive performance” on a variety of tasks.

While Meta has used advances in generative AI and its own investments in technology to bolster its advertising business and increase efficiency across the company, it has yet to significantly crack the AI ​​model market while seeing top competitors in the space advance rapidly. The combined value of OpenAI and Anthropic is now over $1 trillion, and Google’s Gemini technology and services are gaining traction, especially in the consumer market. The risks are huge, as the global generative AI market is estimated to grow more than 40% annually, rising from approximately $22 billion in 2025 to almost $325 billion in 2033. Grand View Research.

Meanwhile, Meta is increasing its investments in AI infrastructure, trying to keep up with spending at other hyperscalers. Meta said in its latest earnings report that AI-related capital expenditures in 2026 will be between $115 billion and $135 billion, nearly double last year’s capital expenditure.

The new Muse Spark will be proprietary rather than open source, and the company said it has “the hope that future versions of the model will be open source.” The company took an open source approach to artificial intelligence with its Llama model family.

Meta said: technical blog About the new model, he noted that improved AI training techniques, as well as a rebuilt technology infrastructure, allow the company to create smaller AI models with “much less computing” capacity than the older mid-sized Llama 4 variant.

“Muse Spark delivers competitive performance on multimodal perception, reasoning, health, and action tasks,” Meta said in the post. “We continue to invest in areas where there are existing performance gaps, particularly long-term agency systems and coding workflows.”

Meta is also experimenting with a new AI model revenue stream by offering third-party developers access to Muse Spark’s core technology via an API. Only unspecified “select partners” currently have access to a “private API preview” of the AI ​​model, but Meta said it plans to offer paid API access to a wider audience at a later date.

The new model now powers the company’s digital assistant on its standalone Meta AI app and desktop website. Muse Spark will be available in the coming weeks on Facebook, Instagram, WhatsApp and Messenger, as well as on the company’s Ray-Ban Meta AI glasses. Meta also plans for Muse Spark to eventually power the Vibes AI video feature in the company’s Meta AI app. This service currently uses AI models from third parties such as Black Forest Labs.

With Muse Spark, users of the standalone Meta AI app and associated website will now be able to switch between specific modes depending on the complexity of their prompts. Users can use one mode that provides quick answers to simple questions and another for more complex queries related to tasks like analyzing legal documents or gathering nutritional information from photos of grocery items.

With Muse Spark, users of the standalone Meta AI app and associated website will now be able to switch between specific modes depending on the complexity of their prompts. With Instant mode, users can get quick answers to simple questions, while Think mode allows them to enter more complex queries for tasks like analyzing legal documents or gathering nutritional information from photos of grocery items.

Meta also said on the technical blog that a Thinking mode will be “gradually rolled out” in the Meta AI app and site for the most complex queries and tasks. The tech blog noted that in this mode, the Muse Spark model leverages a team of AI agents to assist with “parallel reasoning,” thus helping it “compete with the extreme reasoning modes of frontier models like Gemini Deep Think and GPT Pro.”

The Meta AI revamped with Muse Spark will also include a Shopping mode, which the company says can help people buy clothes or decorate rooms.

“Shopping mode leverages the style inspiration and brand storytelling already available in our apps, unearthing ideas from creators and communities people already follow,” Meta said in its blog post.

Commodity shares were up 8% in midday trading Wednesday, fueled by a broader rally on Wall Street.

This is breaking news. Please check back for updates.

WRISTWATCH: Alphabet, Meta, Microsoft are all collapsing as data center spending increases.