Is Muse Spark Meta’s bid to justify AI costs amid ‘superintelligence’ race?
The social media giant said Muse Spark is purpose-built for Meta products and will power a “smarter and faster” Meta AI. Over time, the goal is to unlock “new features” that “cite recommendations and content” that people can share across its Instagram, Facebook and Threads platforms.
“Muse Spark is our most powerful model yet,” the company said via its announcement. “It currently powers the Meta AI app and website and will be rolling out to WhatsApp, Instagram, Facebook, Messenger and AI glasses in the coming weeks. We will also be offering the model in private preview via API to select partners.”
It added: “Our models are scaling predictably. Muse Spark is an early data point on our trajectory, and we have larger models in development.”
Muse Spark is the first in a new series of large language models (LLMs) – referred to as the Muse series – built by Meta Superintelligence Labs, the team it formed last year to turbocharge its AI progress. It also indicates how the company plans to incorporate its social media content into the AI model.
It comes as US technology giants are eager to prove that substantial AI costs are actually paying off. Interest in AI agents in particular is soaring, with industry leaders like Sam Altman referring to them as key to unlocking accelerated growth in AI.
Meta said its goal is to achieve what it calls “personal intelligence” – an assistant able to help anyone, anywhere with “the things that matter most to them”. It described Muse Spark as a deliberate and scientific approach to model scaling and, although small, is fast and capable enough to “reason through complex questions”.
The company added: It is a powerful foundation, and the next generation is already in development. Muse Spark now powers the Meta AI assistant in the Meta AI app and meta.ai, built to support complex reasoning and multimodal tasks.”
New modes and capabilities are being rolled out globally in the coming weeks, Meta said, as it plans to open access to the underlying technology.
“This is only the start. As our models improve, we’ll continue to build safeguards for things like safety and privacy, starting with the strengthened risk framework and other protections we’re sharing today,” the company said.
“The future of Meta AI is rooted in the relationships and context already at the centre of your life. We are building toward personal superintelligence – an AI that does not just answer your questions but truly understands your world because it is built on it.”
The announcement arrives as company chief executive Mark Zuckerberg faces massive pressure from investors to justify significant spending on AI.
Speaking on threads, he said: “We are building products that don’t just answer your questions but act as agents that do things for you.”
Meta also yesterday announced an expanded partnership with CoreWeave, signing a US$21 billion agreement to leverage CoreWeave’s AI cloud platform to scale its inference workloads.
Related stories
Meta outlines roadmap for new generation of in-house AI chips
Nebius inks $27bn AI cloud infrastructure deal with Meta
OpenAI pauses Stargate UK, citing energy costs and regulation
