AI news · October 13, 2026
Meta releases Muse Wonder for personal and professional AI agents
Meta's Muse Wonder is available through its Model API on October 12, with Wonder priced at $4 per million input tokens and $20 per million output tokens; builders can compare it with their current agent model.
- Wonder input tokens
- $4/M
- Wonder output tokens
- $20/M
- models in the release
- 2
Meta released Muse Wonder as two models for personal and professional agent work. Wonder is available through the Meta Model API and Muse Code, while Wonder Full is limited to Muse Code subscribers. Meta says the models are aimed at research, planning, tool use and long-running tasks, with active memory and evaluations that include MemoryBench. The launch puts a broad agent model behind a developer endpoint instead of leaving it as a chat demo.
The price gives builders a useful starting comparison: $4 per million input tokens and $20 per million output tokens for Wonder. Meta reports performance close to Fable 5.1 and GPT-6 Astra on its selected tests, so the numbers need an independent workload check. The practical choice is between this API, a cheaper specialist model and a model already in the stack. Test tool-call success, context recovery and total cost on the same small job before changing a production route.
What you can do with it
Open the Muse Wonder model page and run one agent task with a fixed tool list, a short memory record and a measurable finish condition. Compare completion rate, retries, latency and total tokens with your current model. Keep the API trial small until access, pricing and quality are clear.
Our take
Muse Wonder matters because Meta is offering a serious agent model through its own developer route while also using it to drive a coding product. The release is not proof of a better general model: the strongest comparisons are Meta's, and Wonder Full is gated. A controlled task test will show if distribution and price beat a familiar provider.
Source: Meta / TechCrunch — Made With Models writes the brief; the reporting is theirs.