The race to build more powerful AI keeps getting more interesting, and Meta is now stepping up its game. 

The company has introduced Muse Wonder and Muse Wonder Full, two new AI models designed to compete with leading systems such as ChatGPT and Claude.

So, what makes them different?

An AI that doesn't just answer — it acts

One of the most interesting things about Muse Wonder is that it was designed to work as an AI agent.

Instead of simply answering a question and waiting for the next prompt, it can work through longer and more complex tasks while keeping track of context and trying different approaches when needed.

Meta highlights two features called Active Memory and Active Focus.

Active Memory allows the model to remember important information and bring it back when it becomes useful. Active Focus helps it stay on track during longer tasks, try alternative approaches when something doesn't work, and ask for confirmation before taking important actions.

It can also help build software

Another major focus of Muse Wonder is coding.

Meta says its new models can work on software projects, from smaller applications to more complex development tasks, while keeping track of requirements, making code changes and reviewing the results.

In simple terms, the goal is to move from an AI that tells you how to do something to an AI that can actually help you do it.

And that's where the competition with other AI companies gets really interesting.

Is it going to replace ChatGPT?

It's way too early to say.

Meta says Muse Wonder delivers state-of-the-art performance across different tasks and represents one of its most capable AI models so far. However, performance can vary depending on the task and benchmark being used.

What is clear is that Meta doesn't want to sit on the sidelines of the AI race.

Muse Wonder is already available to developers through Meta's API and Muse Code, while Muse Wonder Full is available through a Muse Code subscription. Meta also plans to bring Wonder into the Muse experience.

What about privacy?

When AI systems can work with personal information and perform tasks on behalf of users, security becomes extremely important.

Meta says Muse Wonder was trained to avoid sharing private information without permission, reject harmful requests and resist manipulative instructions.

Fun fact

Meta says Muse Wonder was trained at a gigawatt-scale data center called Prometheus. ?

Yep, gigawatts.

That gives you an idea of just how much computing power is involved in training and running today's most advanced AI systems.

The race is just getting started

ChatGPT, Claude, Gemini and now Muse Wonder… the competition to build smarter and more capable AI agents is moving fast.

And maybe the most interesting question isn't which AI will win, but how these tools will change the way we work, code, research and solve problems.

Because we're slowly moving from simply asking AI questions to actually asking it to get things done.