Meta has released the Muse Wonder AI models, the latest offering from its superintelligence team following the earlier launch of Muse Spark. The company describes the new models as its most capable for personal and professional tasks, marking what it calls "the next step on our scaling ladder and a major leap toward personal superintelligence."
The Muse Wonder series includes two variants: Muse Wonder and Muse Wonder Full. Both models feature Active Memory and Active Focus capabilities designed to track projects and optimize results for users. Muse Wonder is available through the Meta Model API and Muse Code, while Muse Wonder Full is reserved for Muse Code subscribers.
Performance Comparisons
Meta's models showed competitive performance against rival systems. On the Artificial Analysis Index, Muse Wonder Full scored 56, outperforming OpenAI's Astra model and Anthropic's Claude Fable, though scoring second to Claude Opus.
On the Personal Superintelligence Index, Muse Wonder scored 59.1, performing above both OpenAI and Anthropic's tested models.
Safety Focus
Meta emphasized AI safety measures in the model training process. The company stated it "trained the Muse Wonder family to lead on both dimensions, with robustness to adversarial instructions and careful handling of private information."
Market Reception
Despite Meta's competitive positioning, crypto prediction markets show limited confidence in the company's AI leadership. Data from Polymarket indicates a 3% probability that Meta will have the best AI model by year's end.
