Meta introduced Muse Wonder and Wonder Full, emphasizing Active Memory for retrieving information across conversations and Active Focus for sustaining long tasks. Both target coding, tool use, and professional work and have one-million-token contexts. Official developer documentation confirms API prices for the base model of $4 per million input tokens and $20 per million output tokens, with Full limited to Muse Code subscriptions. Meta claims strong results across several evaluations, but verification effort and costs in actual projects require separate assessment. Availability in the general Muse app is rolling out gradually.