Meta unveils new AI models it says rival OpenAI, Anthropic

Meta unveiled two new "frontier" artificial intelligence models, named Wonder and Wonder Full, on Monday, asserting their performance rivals that of OpenAI's ChatGPT and Anthropic's Claude. Mark Zuckerberg, Meta's CEO, announced on X that Muse Wonder represents a more advanced class of models, positioning it at the forefront. This development follows Meta's US$14 billion investment in San Francisco-based AI startup Scale AI and the hiring of its co-founder, Alexandr Wang, as the company's chief AI officer.

Muse Wonder is slated to power Meta's paid personal agent app, Muse, set to launch soon. Wang declared on X that Wonder is "our largest and most capable model yet" and "rivals the big boys," presumably referring to OpenAI and Anthropic. Meta's blog post highlighted benchmarking data from independent firm Artificial Analysis, indicating that Wonder surpasses previous Meta models and those from Anthropic, OpenAI, and Google.

Despite Meta's hefty AI investments, concerns about safety and security loom large. In July, the company disclosed plans to spend up to US$145 billion on AI-related ventures this year. While Meta has typically moved swiftly, it recently halted the release of Muse for safety reasons. In August, Meta introduced Muse Code to compete with OpenAI and Anthropic's automated coding tools, and in September, launched Muse, an agentic app capable of autonomously performing tasks like coding, research, shopping, and booking reservations.

Meta also unveiled Muse Charm, a handheld AI hardware device enabling users to interact with their Muse agent independently from their smartphone. Despite the industry's focus on agentic apps, there remains debate over the pace of development, with calls for a slowdown to address safety concerns. In September, Zuckerberg signed a "morally binding" commitment alongside six other tech leaders to ensure adequate safeguards around advanced AI models.