Meta has introduced two new artificial intelligence models, Muse Wonder and Muse Wonder Full. The company calls them its most capable models so far and describes them as a step toward what it calls personal superintelligence. The launch puts Meta more directly against OpenAI and Anthropic, at least in the benchmark race.
What the models can do
Meta says the two models add two features it calls Active Memory and Active Focus. Active Memory is meant to carry relevant information across conversations. That could help an AI assistant remember context instead of asking users to repeat themselves. Active Focus is designed to help an agent stay on difficult tasks for longer. Meta did not provide many public details beyond those descriptions, but the features point to a push for agents that can handle multi-step work.
Benchmark scores draw attention
Independent evaluation platform Artificial Analysis scored Muse Wonder Full at 56 on its Intelligence Index. The standard Muse Wonder scored 52. Both scores were recorded at maximum reasoning effort. That places Wonder Full above Claude Fable 5.1 and GPT-6 Astra, which each scored 53, according to the platform. It still trails Claude Opus 5.5, which scored 58. The result is notable because Meta has often been seen as chasing rivals in frontier AI. A score above two major models will give Meta a talking point, though benchmarks are only one way to judge real-world performance.
Pricing and limited access
The new models are Meta’s most expensive to date. The standard Muse Wonder charges $4 per million input tokens. That matches Claude Opus 5.5 and undercuts GPT-6 Astra and Claude Fable 5.1, which charge $10 and $50 respectively. Muse Wonder Full is not open to everyone. It is available only to people who subscribe to Muse Code’s $50 a month power usage tier or its $200 a month maximum usage tier. That makes the top model a premium product for heavy users, not a general release.
Not everyone accepts the claims
Bindu Reddy, CEO of Abacus.AI, challenged the performance claims. Abacus.AI runs its own model evaluations. Reddy wrote on X that Muse Wonder was nowhere near Fable or Astra level. Reddy said detailed benchmark results would follow. Her criticism does not erase Meta’s reported scores, but it shows that model rankings remain contested. Different tests, settings, and prompts can produce different outcomes.
Meta’s launch is a mix of strong benchmark claims, premium pricing, and limited access. The company wants to compete with OpenAI and Anthropic at the top end of the market. Whether Muse Wonder and Muse Wonder Full live up to that ambition will depend on what developers and independent testers find once they use the models in practice.
