Meta is claiming a seat at the frontier-model table. On October 12, the company unveiled two new AI models — Muse Wonder and Muse Wonder Full — saying their coding and agentic capabilities perform at a level comparable to OpenAI's ChatGPT and Anthropic's Claude. CEO Mark Zuckerberg announced the launch on X, while Meta's chief AI officer Alexandr Wang called the flagship "our largest and most capable model yet."

Availability is aimed squarely at developers first. Muse Wonder is accessible through the Meta Model API and Muse Code, Meta's coding assistant, while Muse Wonder Full is available only to Muse Code subscribers. Meta also says the new models will power its Muse personal agent app for paid subscribers — a product that has seen rapid consumer uptake in recent weeks.

The timing matters. Meta spent much of 2024–25 dealing with well-publicized AI setbacks, from leadership turnover to models that underwhelmed against rivals. This launch is the company's first credible claim of parity with the leading labs since that period — and it lands in the middle of an unusually crowded release window that has seen OpenAI's GPT-6 Astra, Anthropic's Claude Opus 5.5 and Google's Gemini 4 Argon all ship within weeks of each other.

As with any frontier launch, the proof will come from independent testing: how Wonder actually benchmarks on coding tasks, what API pricing Meta attaches, and how quickly the Muse app rollout reaches paying users. For now, the claim itself is the story — Meta believes it is back in the race it helped start.

Sources