Meta, formerly Facebook, unveiled Muse Wonder, a closed, paid artificial intelligence model, on the 12th. The move adds a monetization push to a strategy that had centered on free, open AI through its existing Muse service.
Meta released Muse Wonder, its top-tier AI model series, along with a more advanced version called Muse Wonderful. Muse Wonder is offered on a paid subscription basis, with usage priced at $4 per million input tokens and $20 per million output tokens.
While Muse, the AI agent app Meta released earlier, is a personal AI service that offers a free usage allowance, Muse Wonder is the high-performance model that powers it. Meta will give subscribers priority access to Wonder, differentiating its free and paid services by model performance as well as usage limits.
Meta said its models now match the top models from OpenAI and Anthropic on some capabilities. Artificial Analysis, an AI model evaluation firm, scored Muse Wonder at 52 on its intelligence index (AAII), placing it in a tie for seventh behind Anthropic's Claude Fable 5.1, OpenAI's GPT-6 Astra and Google's Gemini 4 Argon, each at 53. Muse Wonderful scored 56, tying for second behind the current leader, Claude Opus 5.5, at 58.
Meta said Muse Wonder includes features such as Active Memory, which retains a user's conversation history, preferences and habits, and Active Focus, which carries complex or ambiguous tasks through to completion.
Meta shares closed the regular session at $742.96, up 3.38% from the previous close, on the strength of the announcement.
In an interview with the Financial Times timed to the launch, Meta Chief Executive Mark Zuckerberg said the company's AI capabilities had caught up with rivals. "We will significantly accelerate the pace of model development," he said. "We are already working on moving to the next stage."
Zuckerberg dismissed concerns that the technology will soon move beyond human control. "My view is much closer to the belief that this is a technology that can be controlled through safeguards and built responsibly, rather than the view that AI will develop a sense of self and act on its own," he said.