Meta Stock Jumps 3.4% as Muse Wonder Takes Aim at OpenAI and Anthropic

Key Takeaways

• Meta stock rose 3.4% to $742.96 after the company introduced Muse Wonder and Muse Wonder Full, its most advanced AI models to date.
• Meta says the models deliver frontier-level performance in AI agents, coding, memory and professional tasks across fields such as finance, law and healthcare.
• The launch strengthens Meta’s AI monetization strategy, but independent testing, user retention and the company’s projected $130 billion to $145 billion of 2026 capital expenditure remain important risks.

Meta Platforms stock rallied on Monday after the company released two advanced artificial intelligence models designed to challenge systems developed by OpenAI and Anthropic.

META shares closed 3.4% higher at $742.96, outperforming a broader technology sector that struggled during the session. The Nasdaq Composite declined approximately 0.4%, making Meta one of the day’s strongest-performing megacap technology stocks.

The gains followed the launch of Muse Wonder and the more powerful Muse Wonder Full. Meta described the models as its most capable systems yet and said they were built for complex personal and professional AI-agent tasks.

The announcement represents a major test of whether Meta can turn its enormous spending on data centers, processors and AI research into products that generate new revenue.

What Are Muse Wonder and Muse Wonder Full?

Muse Wonder is a frontier AI model designed to complete complicated, multi-step tasks rather than simply answer questions or generate text.

Meta says the model can help users interpret leases, navigate insurance claims, analyze financial information and create functioning applications. It is also intended to operate across specialist fields, including law, finance and healthcare.

Two capabilities are central to Meta’s pitch:

• Active Memory: The model can retain relevant information across multiple conversations and retrieve that context when it becomes useful.
• Active Focus: Muse Wonder can continue working when a task becomes difficult, try alternative approaches, recover from errors and request clarification before taking important actions.

These features are intended to make the model more suitable for long-running AI agents. Instead of responding to a single prompt, an agent may need to remember previous decisions, use several applications and adapt when part of a workflow fails.

Meta trained the models at its gigawatt-scale Prometheus data center. The company said Muse Wonder and Wonder Full delivered state-of-the-art results across evaluations covering professional agents, memory and software development.

However, most of the initial performance claims are based on Meta’s own evaluation suite. Independent users and researchers will still need to determine whether those advantages remain consistent in real-world applications.

How Can Developers Access Muse Wonder?

Muse Wonder is available through the Meta Model API and Muse Code, the company’s AI coding platform. Meta has priced API access at $4 per million input tokens and $20 per million output tokens.

Muse Wonder Full is currently available through paid Muse Code subscriptions rather than the general API. Meta is therefore using a tiered model that allows developers to access the standard version on a usage basis while reserving its most powerful system for higher-value subscribers.

The standard Muse Wonder model will also be incorporated into Meta’s consumer-facing Muse agent, beginning with paid subscribers. Meta said it eventually plans to use Wonder to develop specialized models that can be made available to a wider group of Muse users.

The rollout creates several possible revenue streams:

• API fees paid by developers and enterprises
• Muse Code subscriptions
• Consumer subscriptions to the Muse agent
• Transactional revenue when agents complete purchases or bookings
• Greater engagement across Facebook, Instagram, WhatsApp and other Meta applications

Baird analyst Colin Sebastian noted that the models could be monetized directly through subscriptions and API usage and indirectly by keeping Meta’s applications competitive while opening new business opportunities.

Meta Says Muse Wonder Competes With OpenAI and Anthropic

Meta is positioning Muse Wonder against the most capable models from OpenAI and Anthropic.

The company says its new models performed competitively against OpenAI’s Astra and Anthropic’s latest Claude systems on tests involving memory, financial analysis, legal work and long-duration coding. Muse Wonder Full is intended for the most computationally demanding tasks, while the standard model emphasizes a balance between performance, speed and cost.

The release is particularly significant because it marks Meta’s return to the frontier-model competition after a major restructuring of its AI operations.

Meta began overhauling its artificial intelligence strategy approximately 14 months ago. The company formed Meta Superintelligence Labs, recruited prominent researchers and invested heavily in computing infrastructure following concerns that its earlier models were falling behind competitors.

Chief Executive Mark Zuckerberg said the Wonder models demonstrate that Meta has reached the frontier of AI development. The company is also shifting beyond the open-model strategy associated with the Llama family by selling access to a proprietary system aimed at consumers, developers and enterprise customers.

That approach puts Meta into more direct commercial competition with OpenAI and Anthropic rather than limiting its role to distributing models and improving its advertising products.

Why Did Meta Stock Rise?

The market reaction suggests investors see Muse Wonder as evidence that Meta’s expensive AI strategy may be producing commercially competitive technology.

Meta expects capital expenditure of between $130 billion and $145 billion in 2026, including investments in servers, data centers and network infrastructure. The scale of that spending has raised concerns about free cash flow and the time required to generate an adequate return.

A competitive frontier model could help justify those investments by creating revenue outside Meta’s core digital advertising business.

Muse Wonder offers a more visible path to direct AI monetization than improvements to content recommendations or advertising efficiency. Developers can pay for the API, businesses can subscribe to Muse Code and consumers can purchase premium access to the Muse agent.

The models could also improve Meta’s existing products. More capable agents may increase activity across Instagram, Facebook and WhatsApp by helping users shop, communicate, create content and complete administrative tasks without leaving Meta’s ecosystem.

Meta stock has now risen approximately 31% since early September, helped by enthusiasm surrounding Muse and expectations that the company is regaining ground in the AI race. The shares are up about 13% in 2026.

Muse Adoption and Retention Remain Critical

Muse’s early download numbers have been strong, but maintaining engagement may be more difficult than attracting initial users.

The personal AI agent recorded millions of downloads following its September launch. However, reported download growth has since slowed, while retention remains below the level achieved by ChatGPT during a comparable stage of adoption.

Muse Wonder will therefore face an important real-world test: whether better memory, reasoning and task completion can persuade people to use Muse regularly rather than experiment with it briefly.

The agent also depends on access to external websites and applications. Some large retailers, including Amazon and Home Depot, have reportedly restricted Muse’s ability to interact with their websites. Such restrictions could limit the usefulness of shopping and transaction-based agents, particularly if other platforms adopt similar policies. WSJ

Privacy is another risk. An effective personal agent may need access to messages, calendars, documents and financial information. Meta says Muse Wonder was trained to avoid disclosing private information without permission, but the company will still need to earn users’ trust.

Can Muse Wonder Push Meta Stock Higher?

Muse Wonder strengthens the bullish case for Meta by demonstrating that the company can build a competitive AI model and create multiple paths to monetization.

The next stage will depend on execution. Investors will be watching API adoption, Muse subscription growth, user retention and evidence that the models can generate enough revenue to offset Meta’s rapidly expanding infrastructure costs.

Independent performance testing will also matter. Strong internal benchmarks can generate excitement at launch, but developers and enterprise customers will ultimately compare reliability, pricing, latency and security with OpenAI, Anthropic and Google.

For now, the 3.4% increase in Meta stock indicates that investors view Muse Wonder as a meaningful step forward. Continued gains will require Meta to prove that its return to the AI frontier can produce durable revenue growth rather than simply add another layer of spending.

Have a view on where your favourite stocks are heading? Trade selected stock CFDs with Markets.com, with 24/7 trading available on eligible instruments. Take positions on rising or falling prices and explore up to $5,000 in combined rewards for eligible new clients.