Meta is stepping up its artificial intelligence push with two new models designed to compete with leading systems from major technology firms. The company introduced Muse Wonder and Muse Wonder Full on Monday, saying the models can support advanced personal and business AI agents.
The models are available now and form part of Meta’s wider effort to build more powerful AI tools. The company hopes the launch will strengthen its position in a market led by firms such as OpenAI, Google and Anthropic.
The new systems were developed through Meta Superintelligence Labs, which has received major funding from the company. Meta plans to use the models to improve its Muse AI agent, a service launched early last month that has gained popularity.
Meta says the models include two features aimed at making AI assistants more helpful. They are called Active Memory and Active Focus. The tools are designed to help AI systems remember useful details and continue working when tasks become difficult.
Active Memory allows the models to bring relevant information from earlier conversations into new chats. Users may not need to explain their needs again each time they return to the assistant. This could make the service more useful for people who rely on AI for regular tasks at work or at home.
Active Focus is designed to help the models keep working when they face errors or other problems. Rather than stopping immediately, the AI can try to move past a roadblock. Meta also says the models can ask users for more information or permission when needed.
These features are part of a wider industry shift toward AI agents that can take action, manage tasks and respond to changing needs. Technology firms are seeking to move beyond tools that only answer questions. They want systems that can support more complex work with less input from users.
Meta says its new models can match some leading rivals in selected benchmark tests. The company has pointed to Anthropic’s Fable 5.1, OpenAI’s GPT-6 Astra and Google’s Gemini 4 Argon as competitors.
Such claims could help Meta gain attention in the race to develop advanced AI. However, results from selected tests do not show how a model performs in every situation. Independent reviews and real-world use will be important in judging the models’ overall strength.
Meta’s latest launch also signals a change in the company’s position. A year ago, some observers questioned whether it could keep pace with firms that had gained an early lead in generative AI. Its new models and AI products now offer a stronger case for its long-term plans.
The progress has come at a high price. Meta has spent tens of billions of dollars on AI development, infrastructure and hiring. A major investment was its $14 billion deal involving Scale AI, which brought co-founder and chief executive Alexandr Wang into Meta as its chief AI officer.
The company has also recruited researchers and other specialists from rival technology firms. The goal is to build a team capable of developing advanced models and competing at the highest level.
Still, investors have questioned whether the spending will produce enough returns. AI development can require large sums for computing systems, data centres, research and staff. Meta must show that its new tools can attract users and support future revenue growth.
Its recent share performance has trailed some major rivals. The figures cited in the report show Meta shares rising about 6% over the past 12 months. Microsoft gained around 5%, while Amazon rose about 20% and Google climbed roughly 45%.
The figures highlight the pressure on Meta to turn its AI plans into business results. Better technology may improve its products, but investors also want evidence that the investment can strengthen earnings over time.
The company has also introduced its Muse AI agent and plans to launch a device called Muse Charm later this year. These products could help bring its AI services to more users and expand how people interact with its technology.
The next stage will depend on whether users continue to adopt the new tools and whether Meta can develop reliable services at a sustainable cost. The company will need to balance new features with privacy, safety and user trust.
Muse Wonder and Muse Wonder Full give Meta another chance to compete with established AI leaders. Their long-term value, however, will depend on performance beyond benchmark tests and on whether the products deliver measurable gains for the company.
