Meta Platforms CEO Mark Zuckerberg says the company is back at the forefront of artificial intelligence with its latest model, Muse Wonder, as it steps up competition with OpenAI and Anthropic. The launch comes after Meta's Llama 4 model fell short of expectations in April 2025, prompting the company to invest billions of dollars in AI infrastructure, data centres and talent.
In an interview with the Financial Times, Zuckerberg said Muse Wonder would enable “much more sophisticated behaviours that are at the frontier of what is being done in the AI field”.
Released on Monday, the model is designed to handle complex tasks, including interpreting lease agreements and analysing financial and health-related information. According to the Financial Times, Meta claims it outperforms Anthropic's Claude Opus 5.5 and Fable 5.1, as well as OpenAI's GPT-6 Astra, on select technical benchmarks.
Also Read | Layoffs Are Rising In Tech, But These Skills Are Helping People Unlock Better Pay
Meta plans to integrate Muse Wonder into its Muse AI assistant, which uses autonomous agents to perform everyday tasks such as ordering groceries and booking travel. The company says the app has been the most downloaded application on Apple's and Android's US app stores since its September launch.
ALSO READ: Meta's AI Chief Thinks AI Agents—Not Chatbots—Are Gen Z's True Second Brain': Here's Why
Meta shifts towards closed AI models
Muse Wonder marks a shift in Meta's AI strategy. Unlike its earlier emphasis on open-source models, the new system will be closed, with its most advanced versions offered directly to businesses. Pricing ranges from 20 cents per million tokens for basic usage to $200 a month for the premium Muse Wonder Full offering.
Zuckerberg said Meta was already working on its next model, adding that progress in AI development would “probably feel like it'll go quite a bit faster from here”.
Zuckerberg addresses AI safety concerns
As debate over AI safety intensifies, Zuckerberg argued that increasingly capable systems could be developed responsibly with appropriate safeguards.
“I'm much closer to the belief that AI is a technology that can be engineered with normal guardrails and built responsibly than the anthropomorphic vision of AI, which is that it is just going to go do its own thing,” he told the Financial Times.
He declined to disclose his estimate of the risk of AI causing catastrophic harm, known in Silicon Valley as P(doom), arguing that individual AI labs must take responsibility for safety.
Meta is also pursuing Zuckerberg's vision of “personal superintelligence”, focused on empowering individual users.
However, privacy concerns and commercial challenges remain. Amazon has blocked Muse agents from shopping on its platform, while users have questioned how the service handles personal information. Meta has promised a Muse Confidential VM offering designed to prevent the company from accessing users' data and AI conversations.
The company's AI spending has also drawn investor scrutiny. Meta cut around 8,000 jobs, or 10% of its workforce, in May. Zuckerberg said AI could allow companies to expand their operations while employing fewer people, adding that Meta could do “way more things” with “somewhat fewer people”.
ALSO READ: First Came AI Slop. Now AI Slime Mold' is Taking Over the Web
Essential Business Intelligence, Sharp Market Insights, Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
