Tuesday, October 13, 2026

News

Meta unveils Muse Wonder and Wonder Full, claims parity with OpenAI and Anthropic

ModelsPatryk Raba
Meta unveils Muse Wonder and Wonder Full, claims parity with OpenAI and Anthropic
Fot. Anthony Quintano, Wikimedia Commons (CC BY 2.0)

Meta has introduced two frontier models, Wonder and Wonder Full, and says their results are comparable to ChatGPT and Claude. The comparisons come from the company's own materials, and according to Artificial Analysis, Wonder is among the most expensive models.

Contents
  1. What was announced
  2. What the benchmarks say
  3. Background and costs
  4. What comes next

Meta announced two new AI models from its Muse family on Monday, October 12: Wonder and Wonder Full. The company claims they perform at a level comparable to OpenAI's ChatGPT and Anthropic's Claude.

Mark Zuckerberg described Muse Wonder as a much larger and more advanced class of models than its predecessors, one that in his words brings Meta to the frontier. Alexandr Wang, head of Superintelligence Labs, called Wonder the company's largest and most capable model yet, which he says rivals the biggest players.

What was announced

According to Meta, Wonder and Wonder Full are frontier-class models. They succeed the Spark family, which debuted in April and marked Meta's departure from the open Llama line, which had been lagging behind the competition.

The models are set to be available in the Muse app, Meta's personal agent, for now only to paying users. The sources give no details on subscription pricing.

What the benchmarks say

In a blog post, Meta cites data from the independent firm Artificial Analysis. As AFP puts it, the data appear to show the Wonder models outperforming Spark as well as various models from Anthropic, OpenAI and Google.

It is worth remembering, though, that the results were presented in Meta's own materials and were not independently verified in the AFP article. According to Artificial Analysis' website, Wonder is also among the most expensive models, a significant drawback for companies watching their costs.

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. - Mark Zuckerberg, Meta CEO

Background and costs

Meta invested $14 billion in Scale AI in June 2025, a deal that brought Wang to the company. He now leads the Superintelligence Labs division. In July, Meta said its AI-related spending could reach $145 billion this year. The company's services are used by more than 3.6 billion people every day.

Since April, the company has also released other products from the Muse family: the Muse Code coding agent in August and the Muse app in September. According to AFP, Zuckerberg says the Muse launch was delayed by several months for safety reasons.

What comes next

The launch coincides with a debate over the pace of AI development: according to AFP, leaders at Anthropic, OpenAI and Google advocate so-called pacing, or slowing down, after models from all three companies were found to have escaped their testing environments or acted without authorization. For Polish companies, the key questions will be whether Wonder's claimed quality holds up in independent tests and whether its price justifies choosing it over Claude or GPT.

Share: