Meta Launches Muse Wonder AI Models to Catch Up With Rivals

Meta debuted its new Muse Wonder and Muse Wonder Full premium AI models on Monday, priced at up to $200 per month, in a push toward what CEO Mark Zuckerberg recently called the "AI frontier." The launch is the company's bid to catch up with Anthropic, OpenAI and Google, the heavy hitters that have dominated the top tier of artificial intelligence.

Meta calls the new paid tiers its "most capable models to date." The models build on Muse Spark, launched earlier this year, and Meta says they can perform personal and professional agentic tasks. The range spans handling insurance claims, helping renters understand a lease agreement, and building entire working apps.

The stakes are straightforward. Meta has spent billions of dollars over the last year developing these models, and it claims they leapfrog the competition — even as it trails rivals who moved first. The gap between Meta's marketing and its market position is the story behind Monday's announcement.

What did Meta actually release?

Two models, two pricing paths. Muse Wonder is the "state-of-the-art" premium-tier model, available first to coders who pay for access to the Meta Model API. That access comes via the Power Plan at $20 per month or the Maximum Plan at $100 per month. Meta's announcement says the model "will soon power experiences in Muse" with more flexible per-million-token pricing.

Muse Wonder Full is the bigger sibling. Meta describes it as an even more powerful version of the premium Wonder model, with all of the same new features but a shift in focus toward businesses and power users. Wonder Full unlocks more flexible reasoning, a tweaked base model and access to more behind-the-scenes processing power. Meta positions it as a better fit for unfamiliar tasks, engineering and professional workflows.

There is no public API for Wonder Full. It is available exclusively to developers who sign up for the paid Muse Code AI agent, bundled into a monthly subscription priced from $50 to $200.

What can Muse Wonder do that Spark couldn't?

Meta says Muse Wonder builds on the core capabilities of Muse Spark with more advanced features, greater processing power and improved capability in law, finance and health applications. The company's own examples: Muse parsing a leasing agreement or handling an insurance claim. (As the source report notes, do not rely on AI for healthcare or mental health.)

Muse Wonder also brings what Meta calls "frontier coding performance," which the company says will enable the agent to assist in building games and apps.

The release introduces two new features: Active Memory and Active Focus.

• Active Memory enables the model to remember and carry information across conversations. Storing potentially sensitive data raises immediate privacy questions. Meta says it has trained the model to avoid sharing private information without permission and has hardened it against cyberattacks. The company also says "developers and security teams can use models like Muse Wonder to find flaws in their own code and strengthen their systems."
• Active Focus keeps the AI agent working when it runs into technical dead ends. With it, Muse Wonder is said to be capable of alternative approaches, recovering from errors and asking the user for clarification. Meta says that should let the model handle more complex tasks than Spark and encounter fewer roadblocks — something CNET reports seeing firsthand in its testing.

Why does Active Focus raise eyebrows?

The feature's premise — an AI that circumvents blocks in its way — sits uneasily with current policy fights. As the source report notes, boasting that your AI model can more effectively get around barriers that may have been put in place for a reason sounds familiar. Recently, Anthropic revoked access to its AI models, Mythos 5 and Fable 5, after the US government imposed sweeping restrictions on who could use them.

That context matters. The same capability that makes an agent more useful — persistence past obstacles — is the one regulators are watching most closely.

How does Meta's safety stance differ from rivals'?

This launch lands in the middle of an open industry split. Many researchers and experts are advocating for external evaluation and regulation of the technology. Some AI executives, including Anthropic CEO Dario Amodei and OpenAI's Sam Altman, have made high-profile calls to slow AI development.

Zuckerberg has been vocal in the opposite direction. He argues labs can and should self-police to stave off the worst effects of AI.

"I'm much closer to the belief that AI is a technology that can be engineered with normal guardrails and built responsibly than the anthropomorphic vision of AI, which is that it is just going to go do its own thing," Zuckerberg told Financial Times in an interview this week.

That is not an abstract disagreement. Whether Meta's internal safeguards hold up under external scrutiny will shape how quickly enterprise customers — the target of Wonder Full — trust the new models with legal, financial and health-adjacent workflows.

What are experts saying?

"Meta is making up for lost time with lots of new AI tools," Katelyn Chedraoui, CNET's expert reporter on artificial intelligence, told the outlet. "But this new frontier model comes amid serious safety concerns in the AI industry. Safeguards will be essential to any new launch."

Her framing captures the dual reality of Monday's release. Meta is shipping fast and pricing aggressively, with entry at $20 per month. But the frontier-model label invites frontier-level scrutiny, and the industry's own leaders are split on whether any lab should be trusted to police itself.

What comes next?

Meta says per-million-token pricing is coming to Muse Wonder soon, a signal that the company expects usage well beyond the coder early adopters now paying for API access. Whether Muse Wonder and Wonder Full can actually close the gap with Anthropic, OpenAI and Google will depend on the safeguards Chedraoui flagged — and on whether Zuckerberg's bet on self-policing holds up as these agents take on insurance claims, lease agreements and app builds at scale.