Meta Launches Muse Wonder Frontier AI Models, Shares Jump Nearly 5%

Meta Platforms entered the frontier AI race on October 12 with the release of Muse Wonder and Muse Wonder Full, its most capable AI models since the company overhauled its AI division. Meta claims the models deliver performance comparable to OpenAI and Anthropic in coding, financial, and legal tasks, and independent testing cited by Decrypt found the flagship beats GPT-6 Astra and Claude on memory, legal, and marathon coding benchmarks.

The models will power the subscription tier of Meta's Muse AI assistant and are available to developers through the Meta Model API and Muse Code. Investors responded sharply, with Meta shares rising as much as 4.7% on the day of the announcement. Despite the strong internal claims, independent rankings still place Anthropic's Opus 5.5 at the top of the leaderboard, and a Seeking Alpha analysis published the same day found Google holding a significant lead over Anthropic, Meta, and OpenAI for the best model in October.

The launch was also shadowed by a separate report that Meta's Muse AI agent produced explicit content despite a platform ban on adult material, raising compliance concerns. Whether the claimed frontier-level capabilities hold up will depend on forthcoming independent evaluations.