MENLO PARK, CALIFORNIA / RankWire.AI / – Meta has unveiled Muse Wonder and Muse Wonder Full, a new set of sophisticated artificial intelligence models. The company developed these systems to support software development, AI agents, and intricate everyday activities. They are designed to assist with areas such as leases, insurance claims, financial data, and application creation. Meta launched Muse Wonder via its Model API and Muse Code on October 12. Muse Wonder Full became accessible to users with qualifying Muse Code subscriptions.
These models are tailored for multi-step tasks rather than simple, single-response queries. Muse Wonder is capable of planning workflows, utilizing digital tools, and breaking down larger projects into smaller, specialized sub-agents. Additionally, it can operate seamlessly across browsers, desktop applications, and mobile platforms. Meta engineered the system to recover from failures in selected approaches. The models support context windows of roughly one million tokens, enabling them to handle extensive documents, codebases, and other large-scale content within a single session.
Furthermore, Meta has introduced Active Memory and Active Focus features to enhance the Wonder models. Active Memory retains valuable information from earlier interactions and retrieves it when necessary. Active Focus allows an AI agent to persist in solving complex problems and adapt its strategy after errors. The models can ask for clarification if instructions lack critical details and seek confirmation before executing sensitive or significant actions. These features are intended to facilitate longer work periods without users needing to restart tasks repeatedly.
Enhanced agent and coding capabilities with new models
Muse Wonder is priced at $4 per million input tokens via the Meta Model API, with output costing $20 per million tokens, and cached input at $0.20 per million tokens. Instead of separate API pricing, Meta offers Muse Wonder Full through paid Muse Code plans. The company also provides a Power Usage plan at $50 per month, and a Maximum Usage plan costing $200 monthly, aimed at developers requiring greater access for intensive coding tasks.
Meta shared benchmark results that cover areas like software engineering, coding agents, and extended technical tasks. The company highlighted tests such as DeepSWE and SWE-Marathon. The evaluation also included testing under maximum reasoning effort, with Muse Wonder achieving an Intelligence Index score of 52, and Muse Wonder Full scoring 56. These figures place the new models above Meta’s earlier Muse Spark systems on the evaluator’s scale, indicating a clear improvement in reasoning capabilities based on testing results.
Safety protocols and privacy safeguards accompany the launch
Meta emphasizes that the Wonder models incorporate protections for privacy, against harmful requests, and cybersecurity concerns. The models are trained to prevent disclosing private information without user approval. They are also programmed to reject requests involving harmful activities, while still supporting legitimate security functions. Developers are able to utilize the models to identify vulnerabilities in their own software and bolster defenses. Meta assesses its most advanced AI systems under its Superintelligence Scaling Framework and intends to release further details about safety testing for these new models.
This release follows a significant restructuring of Meta’s AI efforts that started approximately 14 months ago. The company states that Muse Wonder builds upon the technical foundations laid by Muse Spark. Development of the new models utilized computing resources from Meta’s Prometheus data center. Additionally, Meta plans to integrate Muse Wonder into more features within the Muse ecosystem, initially targeting subscribers. While developers can currently access Muse Wonder, Muse Wonder Full remains available through qualifying Muse Code subscription plans.