Meta Launches 'Muse Wonder' and 'Muse Wonder Full' with a One-Million Context Window

Meta Superintelligence Labs has launched two new frontier models, "Muse Wonder" and "Muse Wonder Full", designed for personal and professional use via AI agents, operating with a context window of one million tokens, according to the "Artificial Analysis" platform.

The two models scored 56 and 52 points on the platform's Intelligence Index v4.3, compared to 48 points for the "Muse Spark 1.3" model launched six weeks ago, which the platform describes as a substantial leap in performance.

The platform ranked the "Muse Wonder Full" model in the top three across six out of ten evaluation benchmarks, placing it at the heart of the competition for the most capable models.

The "Muse Wonder" model is available via Meta's API, whereas the "Muse Wonder Full" model is limited at launch to "Muse Code" subscribers, a division that reserves the most powerful version for paying customers.

What Do These Terms Mean?

Frontier Model: An AI model from the highest capability tier, used for complex tasks and compared to the most powerful available models.

Context Window: The maximum amount of text or data a model can read and recall at one time, measured in tokens.

Token: The smallest text unit processed by the model, roughly corresponding to part of a word.

API: A channel allowing other software to use the model and send requests to it automatically.