Tuesday 13 Oct 2026
Meta claims new Wonder AI models rival ChatGPT and Claude. (Photo/Reuters)
Meta has unveiled two new "frontier" artificial intelligence models named Wonder and Wonder Full, claiming the systems perform at levels comparable to OpenAI's ChatGPT and Anthropic's Claude as the tech giant accelerates efforts to diversify beyond advertising revenue.
"Muse Spark was our first family of models, and now Muse Wonder is a much larger and more advanced class of models that brings us to the frontier," Chief Executive Mark Zuckerberg wrote on the X platform.
The release marks Meta's latest aggressive push under Chief AI Officer Alexandr Wang, whom Zuckerberg recruited following a $14 billion investment in San Francisco startup Scale AI in June 2025 to head its Superintelligence Labs division.
Wonder, which succeeds the earlier Spark model and departs from the open-source Llama lineage, will soon be available exclusively to paying subscribers on Meta's personal agent application, Muse.
Wang described Wonder as Meta's "largest and most capable model yet" that "rivals the big (boys)".
Safety debates
Independent benchmarking from Artificial Analysis indicates that Wonder outperforms previous Meta systems alongside select models from OpenAI, Anthropic, and Google, whilst ranking amongst the market's most expensive to operate.
Meta, which counts more than 3.6 billion daily active users across platforms including Facebook and Instagram, indicated in July that annual AI capital expenditure could hit $145 billion this year.
The rollout follows Meta's recent launch of Muse Code and the handheld hardware device Muse Charm, designed to capture user demand for autonomous "agentic" software that executes research, bookings, and coding tasks.
However, the rapid rollout comes amid growing industry friction over safety guardrails.
In September, Zuckerberg joined executives from Google, OpenAI, Anthropic, xAI, and Nvidia in signing a White House commitment to implement safety frameworks.
While Meta says it delayed Muse's release to strengthen safeguards, competitors have urged caution after several experimental models escaped secure environments or took unauthorised actions.
___
Source: TRT
