MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has launched its latest advanced AI models, Muse Wonder and Muse Wonder Full, marking the most powerful systems it has developed to date. These new models are tailored for tasks such as coding, managing extended AI agents, and handling intricate personal or professional activities. According to Meta, these models can assist users with insurance claims, lease comprehension, financial analysis, and application development. Muse Wonder was made available on October 12 via the Meta Model API and Muse Code, while Muse Wonder Full was released specifically for Muse Code subscribers.
This launch signifies Meta’s expanded focus on AI systems capable of executing multi-step processes rather than merely responding to isolated prompts. Muse Wonder is designed to plan extended tasks, assign sub-tasks to auxiliary agents, utilize software tools, and recover from errors during operation. Additionally, Meta engineered these models for advanced usage across web browsers, desktop applications, and mobile platforms. Both models support a context window of approximately one million tokens, enabling the processing of substantial data volumes within a single session.
To support longer and more customized interactions, Meta introduced two features named Active Memory and Active Focus. Active Memory maintains relevant information across conversations and retrieves details when necessary for new tasks. Active Focus allows the models to persist with challenging tasks, test alternative strategies, and recover from mistakes. Meta also highlighted that the models can seek clarification from users and verify actions before execution.
Models focus on coding and persistent AI agent tasks
The cost for Muse Wonder is $4 per million input tokens and $20 per million output tokens through the Meta Model API. Cached input data costs $0.20 per million tokens, as per the company’s developer details. Muse Wonder Full is accessible exclusively via paid Muse Code subscriptions, not through separate API pricing. For developers with more intensive coding needs, Meta offers Power Usage at $50 a month and Maximum Usage at $200 a month.
Meta claims that Muse Wonder Full surpasses several agentic and coding benchmarks outlined in the published results. Notably, the tests DeepSWE and SWE-Marathon were used to evaluate software engineering and long-term coding performance. Independent reviewer Artificial Analysis awarded Muse Wonder an Intelligence Index score of 52, while Muse Wonder Full received a score of 56 at maximum reasoning capacity. These scores position the models above Meta’s earlier Muse Spark generation on the evaluator’s index.
Enhanced safety features accompany expanded model functionalities
Meta has trained the Muse Wonder series to refrain from sharing private data without permission and to decline harmful requests. The models also retain cybersecurity functions designed to help developers identify vulnerabilities in their code and bolster security measures. Meta states that the models are assessed under its Meta Superintelligence Scaling Framework. Furthermore, the company plans to release a comprehensive report on additional safety evaluations following the initial rollout.
Approximately 14 months prior to launching Muse Wonder, Meta initiated a broad revision of its AI initiatives. The company explains that these new models build upon the foundation laid by Muse Spark, with training conducted at its Prometheus data center. Meta intends to integrate Muse Wonder into more features within its Muse ecosystem, beginning with subscribers. The October release provides immediate access to Muse Wonder for developers, reserving the larger Muse Wonder Full model for Muse Code members.