Meta Stock Jumps as New AI Models Put Its Comeback to the Test

Meta Platforms, Inc. rose about 4% on Monday, October 12, 2026, after launching Muse Wonder and Muse Wonder Full, which the company calls its most capable AI models to date. Shares traded around $748 to $749 in the afternoon, up from Friday's close of $718.67, and snapped a four-session losing streak even as the Nasdaq Composite declined. The release is the first real test of whether Mark Zuckerberg's multibillion-dollar rebuild of Meta's AI organization has closed the gap with Anthropic and OpenAI, after the disappointing reception of Llama 4 in 2025. The stock has already rebounded sharply on the Muse AI agent, which launched in September, so expectations going into this release were high. This article looks at what Meta announced, how the models stack up on the company's own benchmarks, what the chart is saying and what could go wrong ahead of the third-quarter report on October 28.

 

 

On Monday, October 12, Meta announced Muse Wonder and Muse Wonder Full, a new family of models from Meta Superintelligence Labs, the unit led by chief AI officer Alexandr Wang. Zuckerberg announced the family on X, describing it as a set of frontier models with state-of-the-art performance on personal and professional agentic work. In a blog post, Meta said it began overhauling its AI efforts 14 months ago and has gone from rebuilding its foundations to pushing the frontier. The models were reportedly developed under the internal codename Watermelon.

The product plan is specific. Muse Wonder will soon power the Muse personal AI agent, initially for paying subscribers on the $20 and $100 tiers. The more powerful Muse Wonder Full is available to subscribers of Muse Code, Meta's AI coding tool. Meta said the models are built for tasks such as understanding a lease or making sense of financial or health information, and that they were trained to avoid sharing private information without permission.

The release follows a rapid sequence of launches. Muse Spark, the first model from Superintelligence Labs, debuted in April and was a closed model, a break from the open-weight Llama approach. Muse Image followed in July. In September, Meta launched the Muse agent, then showed new AI glasses and a keychain-sized device called Muse Charm at its Connect event on September 23. On September 28 it created the Meta Enterprise Platform, which brings Muse, Meta Business Agent, the Muse API and Muse Code together under former MongoDB CEO CJ Desai, who reports to Zuckerberg.

The benchmark table Meta posted is the most scrutinized part. According to a count by FourWeekMBA of the table attached to Zuckerberg's post, Muse Wonder Full had the top score on six of 14 rows and Muse Wonder on one, while Anthropic's Claude Opus 5.5 led on six and one went to a rival OpenAI model. On the Artificial Analysis Intelligence Index row, Claude Opus 5.5 scored 58, Muse Wonder Full 56, Claude Fable 5.1 and an OpenAI model 53, and Muse Wonder 52. These are Meta's own figures, and independent results will matter more.

 

The immediate reason is simple: investors wanted evidence that Meta's AI spending is producing a competitive product. Bloomberg described the launch as a milestone that tests whether the multibillion-dollar rebuild has improved Meta's technology enough to compete among leaders. A model that places second on one widely watched index, and leads on several benchmark rows, is a much better answer than the Llama 4 reception Meta had to live down.

The stock also had room to bounce. Before Monday it had lost four sessions in a row, closing at $718.67 on October 9, about 8% below its September 24 high of $779.82. That level followed a surge of about 27% in September, from $572.34 at the start of the month, driven by the Muse agent launch on September 8. By October 1 the stock had closed at $725.93, about 7% below the peak, and it traded near $741 on October 6 as investors weighed the enterprise push. Monday's gain brought it back to roughly $748 to $749.

Analysts have been supportive. Wells Fargo raised its price target to $1,000 on October 6, JPMorgan reaffirmed an Overweight rating and lifted its target to $920 from $820, and Citigroup expects the Muse assistant could generate more than $27 billion of annual revenue by 2030. Cantor Fitzgerald's Deepak Mathivanan kept an Overweight rating but trimmed his target to $840 from $860 ahead of earnings. The consensus view is broadly positive, with a 12-month average target of about $799 from 63 analysts and a Strong Buy rating on average.

 

The chart shows a sharp V-shaped recovery that has stalled below the high. The stock rose from $572.34 on September 1 to $779.82 on September 24, then pulled back to $718.67 by October 9. Monday's close near $748 sits in the middle of that range. The 52-week range runs from $520.26 to $779.82, so the stock trades well above its yearly low and about 4% below its record area.

The levels that matter are clear. The $779.82 high is the first resistance, and a close above it would confirm the uptrend has resumed. The $741 area, where the stock traded on October 6, and the October 1 close of $725.93 are the first support levels, with Friday's close of $718.67 below them. A close under $718.67 would erase Monday's reaction and suggest the launch was sold.

 

 

After a 27% monthly gain, the RSI(14) was likely elevated in September and has cooled during the pullback, which is a healthier setup than chasing a vertical move. The 50-day simple moving average is the trend reference to watch if the stock retests the lower part of the range.

 

The comeback case says Meta has finally produced a credible frontier model and has the distribution to monetize it. Muse reached about 3 million weekly users and 1 million daily prompt users, and third-party data shows a seven-day retention rate above 30%, roughly on par with ChatGPT, and 2.8 million downloads in the first two weeks, trailing only ChatGPT. JPMorgan cited the growth of Muse Connectors and long-term monetization potential. The core business is also strong: second-quarter revenue grew 28% to $60.80 billion, with advertising up 27%, and the stock trades at about 28 times earnings, which Motley Fool called attractive for a company at the front of AI. Meta's own benchmark table puts Muse Wonder Full within two points of the top model on a major index.

The unproven case starts with what Meta did not claim. Muse Wonder Full ranked second on the index in Meta's own table, not first, and the numbers come from the company's benchmark selection rather than independent testing. Monetization is still early: Benzinga reported that Muse downloads are slowing and that Hunterbrook Capital has taken a short position on the idea that early excitement is running out of steam, and BNP Paribas analyst Nick Jones has said the product still needs to prove it can generate revenue. Amazon has blocked the Muse agent from its platform, which limits what the agent can do for shoppers. The spending is enormous, with capital expenditures cited around $145 billion, and Meta repurchased no shares in the first half of 2026 versus $22.92 billion a year earlier. Add to that a New Mexico jury verdict and a pending privacy penalty, a UK Ofcom investigation, and a planned legal charge of roughly $10 billion in the third quarter, and bears have plenty to point to.

 

The nearest risk is the October 28 earnings report. Meta typically moves sharply on results, and this report carries three unusual items: a legal charge of about $10 billion, first guidance on 2027 capital spending, and early metrics on Muse monetization. A stock that has already rallied on the AI story can sell off if capex guidance disappoints or if subscription numbers lag. One Motley Fool columnist has argued the stock could drop after earnings and that this could be a buying opportunity, which illustrates how divided opinion is.

There are also execution and regulatory risks. If Muse Wonder's real-world performance does not match Meta's benchmark table, the narrative could reverse quickly, and the whole Muse effort is being measured against the standard set by Anthropic and OpenAI. Insider selling has been a minor headwind, including a planned sale of 27,474 shares worth about $21.4 million by entities linked to Zuckerberg on September 24 and 40,000 shares by Chris Cox on September 21, though such sales under pre-arranged plans are common. Meta has also barred ByteDance and TikTok from buying ads on Facebook and Instagram in seven countries, a reminder that the ad business carries its own geopolitical exposure. For risk management, traders may consider using Friday's close of $718.67 as a reference for where the launch reaction would be fully reversed, and sizing for a stock that can move 5% to 10% on earnings.

 

Meta's launch of Muse Wonder and Muse Wonder Full gives the company something it lacked for most of the past year: a model that can plausibly be described as frontier-class. The 4% jump to about $748 shows investors are willing to give Zuckerberg credit, and the 27% September rally shows how much belief is already priced in. The test from here is whether the benchmarks hold up outside Meta's own table, whether subscribers pay for Muse, and whether the earnings report on October 28 shows spending that can be justified by revenue. With a $779.82 high just overhead and a $10 billion legal charge coming, do you think Meta's comeback is real enough to carry the stock to new highs, or has the market already paid for it?

 

A: Meta launched Muse Wonder and Muse Wonder Full, which it calls its most capable AI models yet. Muse Wonder will power the Muse personal AI agent for subscribers, and Muse Wonder Full is available to subscribers of the Muse Code coding tool.

A: META rose about 4% to roughly $748 to $749 in afternoon trading, versus a Friday close of $718.67. The gain ended a four-session losing streak.

A: In Meta's own benchmark table, Muse Wonder Full scored 56 on the Artificial Analysis Intelligence Index versus 58 for Anthropic's top model, and led on six of 14 benchmark rows. These are company-published figures, so independent testing will be the real check.

A: Meta is expected to report third-quarter results on October 28, 2026, after the U.S. close. The report is expected to include a legal charge of about $10 billion and will show how much the company plans to spend in 2027.

A: Analysts are broadly positive. Averages put the 12-month price target near $794 to $799, with a Strong Buy consensus, while Wells Fargo is at $1,000, JPMorgan at $920 and Cantor Fitzgerald at $840.