Jakarta, INTI - Meta has officially introduced two new artificial intelligence (AI) models, Muse Wonder and Muse Wonder Full, marking the company’s first frontier models designed to handle complex tasks, including software programming.
According to a TechCrunch report published on Monday, October 12, 2026, both models were developed by Meta’s Superintelligence Lab and offer enhanced capabilities compared with Muse Spark, the company’s earlier lightweight AI model designed for faster response times.
“Together, these models represent the next step in scaling our models and a major leap toward personal superintelligence,” Meta stated in its official announcement.
The launch marks Meta’s latest effort to strengthen its position in the advanced AI market, competing with technology companies such as OpenAI and Anthropic.
Muse Wonder Claims Capabilities Comparable to GPT-6
Meta claims that Muse Wonder and Muse Wonder Full feature larger architectures and improved performance on complex tasks compared with previous generations of its AI models.
According to the company’s internal evaluations, both models achieved benchmark scores comparable to Anthropic’s Fable 5.1 and OpenAI’s GPT-6 Astra.
Their capabilities include tasks requiring advanced reasoning and software programming.
Meta has set the pricing for Muse Wonder through the Meta Model API at US4permillioninputtokensandUS20 per million output tokens.
These rates are broadly comparable to those of OpenAI’s GPT-6 Sol and Anthropic’s Opus 5.5.
Meanwhile, Muse Wonder Full is available exclusively to Muse Code subscribers and cannot be accessed through the API.
Both models will also be used to enhance the capabilities of Muse, Meta’s AI agent.
Meta Introduces Long-Term AI Memory Capabilities
Beyond handling complex tasks, Muse Wonder and Muse Wonder Full feature active memory technology designed to retain and consolidate user information over extended periods.
The feature aims to enable AI systems to remember relevant details about users for days, weeks, or even months.
This capability is intended to support the development of personalized AI assistants that can manage ongoing tasks and deliver responses informed by previous interactions.
To evaluate this functionality, Meta used MemoryBench, a benchmark designed to assess continual learning and memory formation in AI systems.
According to Meta’s internal testing, Muse Wonder achieved a higher MemoryBench score than the Fable, Opus, and Astra models.
“We’ve invested an enormous amount of effort into ensuring that models can remember the important things about you for days, weeks, months, and beyond,” said Meta researcher Kavya Kopparapu.
Muse Wonder Developed Using the Prometheus Data Center
Muse Wonder and Muse Wonder Full are Meta’s first frontier AI models since Alexandr Wang, head of the company’s Superintelligence Lab, overhauled its AI development strategy in August 2025.
Following the restructuring, Meta first introduced Muse Spark as a lightweight AI model and Muse as a personal AI agent.
At the time, however, the company had yet to introduce frontier models positioned to compete directly with flagship offerings from OpenAI and Anthropic.
Meta said its latest models were trained using infrastructure at the Prometheus data center in Ohio, United States.
Prometheus began operations in 2026 and serves as part of Meta’s infrastructure for developing next-generation artificial intelligence technologies.
Conclusion
Meta has introduced Muse Wonder and Muse Wonder Full, its latest frontier AI models designed for complex reasoning, programming, and personalized assistance. With long-term memory capabilities and enhanced performance, the models reflect Meta’s efforts to compete with OpenAI and Anthropic while advancing its vision of personal superintelligence.
Read more: As AI Management Grows More Complex, Indosat Integrates 300 Models into a Single Platform