Meta Enters Frontier AI Territory With Muse Wonder Models

Meta has launched Muse Wonder and Muse Wonder Full, its first frontier-class models, targeting the high-end reasoning market with pricing competitive to GPT-6 Sol and Opus 5.5.

AI & Culture Desk
Updated 2d ago 2 min read
Meta Enters Frontier AI Territory With Muse Wonder Models

Meta has officially entered the frontier AI market with the release of two new models, Muse Wonder and Muse Wonder Full. According to TechCrunch, these are the company's first attempts at developing state-of-the-art, high-compute models. The rollout follows a period where Meta focused on its lightweight Muse Spark model and Muse agent, avoiding the high-stakes releases typical of competitors like OpenAI and Anthropic. The new models were trained at Meta's Prometheus data center in central Ohio, a facility that only came online earlier this year.

The company is positioning these models as direct competitors to established industry leaders. Meta claims that Muse Wonder and Muse Wonder Full achieve benchmark scores comparable to GPT-6 Astra and Fable 5.1. While Meta has touted these results, the company has not provided independent verification beyond its own internal testing. The models are currently available to Muse Code subscribers and are accessible via the Meta Model API, where they are priced at $4 per million input tokens and $20 per million output tokens, placing them in the same cost tier as GPT-6 Sol and Opus 5.5.

A primary differentiator for the new models is Meta's focus on what it terms active memory. Research scientist Kavya Kopparapu stated in a launch video that the company invested significant effort into ensuring the models can retain and synthesize user information over extended periods. To validate this, Meta introduced the MemoryBench benchmark, which evaluates memory construction and continual learning. According to Meta's internal data, Muse Wonder currently outperforms Fable, Opus, and Astra on this specific metric.

The launch represents a strategic pivot for Meta’s Superintelligence Labs division, which underwent a significant restructuring in August of last year under the leadership of Scale AI founder Alexandr Wang. Since that reorganization, the company has steadily expanded its AI portfolio, including the Muse Image model and its agentic efforts. Meta has indicated that it plans to eventually integrate the Wonder models into its Muse AI agent, though the company has not provided a specific timeline for this transition or detailed how it will scale these capabilities to a wider user base.

While Meta is aggressively attempting to close the gap with OpenAI and Anthropic, the competitive landscape remains fluid. The Verge notes that it is currently unclear how these new models compare to Google’s Gemini 4 Argon, as the capabilities of that system are still emerging. For now, Meta’s entry into the frontier tier provides enterprise customers with an alternative for heavy computational tasks, though the company’s long-term success will depend on whether these benchmark gains translate into practical utility for users managing complex, long-term information.

Sources

  1. 01 Meta releases its first frontier model, Muse Wonder — TechCrunch — AI
  2. 02 Meta has some Wonder Full new AI models — The Verge — AI