Meta has launched two artificial intelligence models, named Muse Wonder and Muse Wonder Full, developed to execute complex tasks — including coding — and retain user information over extended periods of time.
The models were unveiled earlier this week by the Superintelligence Lab, the division of the company dedicated to AI development. According to Meta, both solutions achieve results comparable to Anthropic's Fable 5.1 and OpenAI's GPT-6 Astra.
Muse Wonder has been designed to prioritise speed and efficiency, while Muse Wonder Full represents a larger version with more advanced capabilities. Both were built to support the completion of complex tasks and to adapt to user needs over prolonged periods.
One of the key features of the new models, according to the company, is their ability to retain relevant information across days, weeks, and months, enabling continuity in tasks and allowing responses to be tailored to each user's specific context.
"We put a tremendous amount of effort into ensuring that your model can remember what matters to you across days, weeks, months and beyond," said Kavya Kopparapu, one of the lead developers behind the models, as quoted by TechCrunch.
The new systems are the most advanced models released by Meta's AI division since it came under the leadership of Alexandr Wang, founder of Scale AI.
In a statement accompanying the launch, the company said that Muse Wonder and Muse Wonder Full represent a new step in the development of AI systems capable of delivering more personalised assistance to users.
Source: Diário Económico
Original article: https://www.diarioeconomico.co.mz/?p=552291
