POPA ReconMeta Muse Wonder
Meta describes Muse Wonder and Muse Wonder Full as its newest frontier models for personal and professional agent tasks. Meta says both are available today through the Meta Model API and Muse Code, while Wonder Full is available to Muse Code subscribers. Meta also says Wonder will soon power its Muse personal-agent product, rolling out first to subscribers (Meta AI).
Benchmark claims
The linked posts report that Wonder Full leads on DeepSWE and scores 57.5% on SWE-Marathon, compared with 46.3% for Claude Opus 5.5 and 42.5% for GPT-6 Astra (Kimmonismus). These are reported benchmark results, not independent confirmation. Meta’s Alexander Wang says the models build on more than a year of research and infrastructure work and continued scaling of pre-training and reinforcement learning (Wang).