Meta launches Muse Wonder, a frontier model built for long-running AI agents
Meta released Muse Wonder and Muse Wonder Full on October 12, calling them its most capable models yet for personal and professional agent work. Muse Wonder is available through the Meta Model API and Muse Code at $4 USD per million input tokens and $20 per million output tokens. Muse Wonder Full is only on Muse Code's Power and Maximum Usage plans. Meta highlights Active Memory, which carries relevant context across conversations, and Active Focus, which keeps the model working through blockers, trying other approaches and checking in before important actions. Both models support long-running coding agents, tool use and multimodal inputs such as screenshots and documents.
What it means for manufacturers and fabricators
Another lab is pitching models that can hold a job across many steps and ask before they do something consequential, which is the pattern that matters for office workflows like quoting follow-ups, PO checks and exception sorting, so it is worth watching whether Muse Wonder agents hold up on a real multi-step shop task compared with Claude or GPT-6.
Related guide: What Is an AI Agent? A Plain-English Guide for Manufacturers