Meta prices Muse Wonder like Anthropic's frontier API
In brief
According to Meta, on 12 October it launched two general-purpose models, Muse Wonder and Muse Wonder Full, aimed at long-running tasks and agents. Meta says their Artificial Analysis scores sit near Claude Opus 5.5 and GPT-6. Access is through the API or a Muse Code subscription; weights cannot be downloaded or run locally.
Source : Next - « Meta revient dans la course aux modèles frontière avec Muse Wonder »
What it changes for an SME
For an SME, this adds another frontier-model API at the prices Meta sets in line with Anthropic's Claude Opus 5.5: 4 dollars per million input tokens and 20 dollars per million output tokens. The Full model is limited to Muse Code subscribers on the heavy-use plan, from 42.99 euros a month in France. You cannot install the models locally. Several of the stronger claims on coding, memory and agents come from Meta's own tests.
What you can do
- Compare these API rates with your current Anthropic or OpenAI invoices.
- Check whether the 42.99 euro Muse Code heavy-use plan in France fits your actual volume.
- Trial the models on your own workloads before you switch providers.
Let's talk about the time you could save.
In 20 minutes, we find where AI could save you time or money. Free, no commitment.