Published October 13, 2026 · AI News
Meta has released Muse Wonder and Muse Wonder Full, the first models from its Superintelligence Lab that it calls frontier-level. Meta says they match the best models from OpenAI and Anthropic on coding, legal and financial tasks. For now they are aimed at developers, not at the Meta AI chat most people use.
What Meta released
| Muse Wonder | Muse Wonder Full | |
|---|---|---|
| Where you get it | Meta Model API and Muse Code | Muse Code subscribers only (not in the API) |
| API price | US$4 per million input tokens, US$20 per million output | Not sold through the API |
| Built for | Agents, coding, law, finance, health questions | The same, with more compute |
| Coming next | Will power more of the Muse app, subscribers first | — |
Two features get most of Meta’s attention. Active Memory lets the model carry what matters to you across conversations. Active Focus is meant to keep an agent working on a hard task, retry other approaches and check in before doing anything important.
How good is it?
Meta’s own testing puts Muse Wonder level with the top models from OpenAI and Anthropic, and ahead of them on a memory benchmark. Treat that with care: the numbers come from Meta, and independent tests aren’t out yet. A full safety report is “coming soon”, according to Meta.
At US$4 / US$20 per million tokens, Wonder is priced in the same range as the flagship models from OpenAI and Anthropic. It is not a budget option.
What it means for you
If you use Meta AI in WhatsApp, Instagram or Facebook, nothing changes today. Muse Wonder reaches paying Muse subscribers first, and Meta hasn’t said which countries get it, so Canadian availability is unclear. Developers can try it through the API now.
Our take: this is Meta catching up, not leaping ahead. For everyday writing, research and coding, ChatGPT, Claude and Gemini remain the safer picks until independent results arrive. Compare them in our best AI chatbots and best AI for coding guides.