Pomegra
Meta shares rose about 4% on Monday after it launched Muse Wonder and Muse Wonder Full, its first frontier AI models, to rival OpenAI and Anthropic.
- Meta launched Muse Wonder and Muse Wonder Full on Oct. 12, its first frontier-class models from Meta Superintelligence Labs.
- Meta Platforms (NASDAQ: META) gained about 4% in afternoon trading on the release.
- Muse Wonder costs $4 per million input tokens and $20 per million output tokens, in line with rival flagships.
Lead
Meta Platforms (NASDAQ: META) launched Muse Wonder and Muse Wonder Full on Monday, Oct. 12. The company called them its most capable AI models to date and its first to compete directly with the leading systems from OpenAI and Anthropic. Shares rose about 4% in afternoon trading. The launch follows a year in which CEO Mark Zuckerberg committed tens of billions of dollars to AI infrastructure and recruited researchers from rival labs.What Is Muse Wonder and How Is It Offered?
Muse Wonder is a pair of multimodal models developed by Meta Superintelligence Labs and trained at the company's Prometheus data center in Ohio. Both accept text, images, audio and video, return text, and carry a context window of roughly one million tokens.
Access is split by product. Muse Wonder is available through the Meta Model API under the identifier muse-wonder-1.0. It costs $4 per million input tokens and $20 per million output tokens, with cached input at $0.20 per million tokens. Muse Wonder Full is not offered through the API. It is limited to subscribers of Meta's Muse Code coding tool, whose plans cost $50 and $200 a month. Neither model offers downloadable weights, a departure from the open-weight approach Meta used for earlier Llama releases.
Meta said Wonder will also power its Muse personal agent, initially for subscribers. The company highlighted two features. "Active Memory" carries context across conversations over days and months. "Active Focus" lets the model pursue a goal through obstacles, revise its plan and ask for user confirmation before consequential actions.
How Does Muse Wonder Compare With OpenAI and Anthropic?
Muse Wonder Full roughly matches the leading commercial models on third-party benchmarks, but it does not lead them. On the Artificial Analysis Intelligence Index, Wonder Full scores 56, level with Anthropic's Claude Sonnet 5.5 and below Claude Opus 5.5 at 58. Wonder scores 52, just under OpenAI's GPT-6 Astra and Google's Gemini 4 Argon, both at 53. Wonder Full posts 79.0 on a software engineering test and 57.5 on long-running coding tasks. Meta also reports higher scores than competing models on MemoryBench, a test of continual learning.
API pricing sits close to that of GPT-6 Sol and Claude Opus 5.5. Meta is therefore competing on persistent memory, agent behavior and distribution across its consumer apps rather than on price.
Why Did Meta Shares Rise?
Meta shares rose because the launch gives investors evidence that the company's AI spending is producing competitive products. The company has guided 2026 capital expenditure to as much as $145 billion, a figure that unsettled investors after its April earnings report. A frontier-class model, a paid coding product and a subscription agent also point to revenue beyond advertising.
The move extends a rally in ai stocks tied to Meta's Muse products. The standalone Muse agent topped U.S. app charts in September, and the shares gained more than 11% in a single session at that time, their largest one-day rise since April 2025. Monday's gain adds to that recovery toward the summer 2025 highs.
Strategic Context
The release closes a gap that opened in 2025, when Meta's Llama 4 models drew criticism and the company reorganized its AI work under Scale AI founder Alexandr Wang. Muse Spark, a lighter model, came first, followed by the Muse agent and now the Wonder family.
The shift to closed weights and subscription pricing marks a change in business model. Meta is moving from distributing models freely to monetizing them directly, which puts it in the same commercial position as OpenAI and Anthropic. It also raises the question of how quickly paid usage can offset infrastructure costs on this scale.
Outlook
Meta has reached the frontier tier, though not the top of it, and the market rewarded the credibility. The next tests are adoption of Muse Code, retention in the Muse agent and the pace of Wonder's integration into consumer products. Rival labs are expected to keep releasing new models, so Meta's position will be measured against successor versions rather than against today's benchmarks. Capital spending guidance and subscriber figures in the next quarterly report will show whether the launch converts into revenue.
