Meta launches Muse Wonder models for agent tasks

Meta announced Muse Wonder and Muse Wonder Full on October 12, describing them as models for coding and personal and professional agent tasks. The company says Muse Wonder is available through the Meta Model API and Muse Code, while Wonder Full is available to Muse Code subscribers. ^[1]

Meta says the models include “Active Memory” to carry relevant details across conversations and “Active Focus” to keep working through difficult tasks. These are company descriptions of intended capabilities, not independent findings about how well the models perform in everyday use. ^[1]

TechCrunch reported that Meta’s own testing showed benchmark scores comparable to certain competing models. The two sources do not provide independent performance assessments; Meta said a fuller safety evaluation report would be released later. ^[2][1]