Meta Platforms Stock Rises as Muse Spark 1.1 AI Model Debuts

Key Points

• Meta Platforms shares have risen sharply in recent days, driven partly by the release of its new Muse Spark 1.1 AI model and cloud computing plans.
• Muse Spark 1.1 scored higher than Alphabet's models on the Intelligence Index but still trails top offerings from Anthropic and OpenAI.
• Meta will charge per-token fees for Muse Spark 1.1 at roughly 25% of rival pricing, marking its first real push to monetize AI models.

After being down and out for several months, shares of Magnificent Seven giant Meta Platforms NASDAQ: META are starting to get their groove back. The stock recently popped 8.8% on reports that Meta will enter the cloud computing business, selling excess capacity to third parties.

Meta shares then saw two large single-day up moves on July 9 and July 10, rising 4.7% and 6%. Part of the stock’s latest gain is due to what is likely Meta’s most significant artificial intelligence (AI) model release: Muse Spark 1.1.

Data indicates that Muse Spark 1.1 is Meta’s most intelligent model yet.

Additionally, the model may mark the beginning of an inflection point in Meta’s ability to generate revenue from AI products.

Muse Spark 1.1: Meta’s AI Model Intelligence Is on the Rise

Artificial Analysis is a helpful source for gauging the relative capabilities of AI models. The company tests models on agentic, coding, general intelligence, and scientific reasoning to give them an “Intelligence Index” score. Currently, Muse Spark 1.1 has a score of 51, which is higher than any model developed by Alphabet NASDAQ: GOOGL. However, it still ranks below many of Anthropic's and OpenAI’s latest models, several of which have scores above 55.

Muse Spark 1.1’s score is also significantly higher than that of Meta’s initial Muse Spark model, with a score of 43. Going forward, it will be important to see whether Muse Spark 1.1 maintains this score and its relative standing among other models. Initially, Meta’s first Muse Spark model had a score of 52, but this has since fallen. This is likely because Artificial Analysis updates and reweights its evaluation framework over time.

Still, based on the latest testing, Muse Spark 1.1 represents a significant improvement over the original Muse Spark and ranks highly overall. This lends validation to Meta’s massive AI capital expenditures and its hiring of Chief AI Officer Alexandr Wang, who has been critical to Muse Spark’s development. Not only is Meta making better models, but for the first time, it is making a real monetization push.

Meta Steps Into AI Model Monetization

Notably, Muse Spark 1.1 marks the first time Meta will charge for access to its models. Meta will charge users on a per-token basis, or based on the amount of information the model processes and outputs, in a "pay-as-you-go" format. Anthropic and OpenAI allow users to pay for models in this way as well, but also provide access through flat monthly or annual fees.

One of the reasons to think that Muse Spark 1.1 could gain real traction and generate notable revenue for the firm is its pricing. CEO Mark Zuckerberg says Muse Spark 1.1’s per-token pricing is around 25% of what Anthropic and OpenAI charge for similar models. Artificial Analysis adds weight to this. It places Muse Spark 1.1’s “cost per Intelligence Index Task” around three times lower than OpenAI’s GPT-5.4, which also has an Intelligence Index Score of 51.

If Muse Spark 1.1 offers a level of intelligence comparable to another model but at a much lower cost, users have an incentive to adopt it. This gives Meta a realistic opportunity to start generating significant revenue directly through its AI model. This may come through software developers using it for coding tasks, an area where its performance is particularly improved over the original Muse Spark.

Still, it is possible that Meta is highly subsidizing its model cost, with Alexandr Wang calling the pricing “very aggressive and attractive.” The word ‘aggressive’ seems to indicate a degree of deliberate undercutting. In turn, Meta’s pricing may not be high enough to support the model profitably.

Nonetheless, through low pricing, Meta has an opportunity to prove Muse Spark’s capabilities to users, an important first step in generating sales. Over time, Meta can evolve its pricing to increase margins.

Patience Remains Key as Meta Looks to Monetize Muse Spark 1.1

There is real reason for investors to feel excited about the progress Meta has made with Muse Spark 1.1. It has a better model and is now looking to monetize it to boost returns on its AI spending. Still, the true test will be what Meta shows over time in its actual financials.

Investors should monitor the company’s future earnings calls for data on how much revenue Muse Spark is bringing in. It may take time for Meta to provide detailed information on this, and the company’s near-term earnings reports may not give much insight.

Meanwhile, recent gains indicate investor optimism, and sentiment around Meta has not been this high in quite some time.