Meta delays Avocado AI model rollout to May after falling short against Google and OpenAI

Meta Delays Rollout of New A.I. Model After Performance Concerns

The tech giant pushed back the timeline after spending billions to be on the cutting edge of artificial intelligence. Mark Zuckerberg, the chief executive of Meta, said in July that his company's new artificial intelligence models would "push the frontier in the next year or so." Now Mr. Zuckerberg -- who has invested billions in the A.I. race -- appears increasingly unlikely to hit that deadline, three people with knowledge of the matter said. Meta's new "frontier" A.I. model, which the company has been working on for months, has fallen short of the performance of leading A.I. models from rivals like Google, OpenAI and Anthropic on internal tests for reasoning, coding and writing, said the people, who were not authorized to speak publicly about confidential matters. The model, code-named Avocado, outperformed Meta's previous A.I. model and did better than Google's Gemini 2.5 model from March, two of the people said. But it has not performed as strongly as Gemini 3.0 from November, they said. As a result, Meta has delayed Avocado's release to at least May from this month, the people said. They added that the leaders of Meta's A.I. division had instead discussed temporarily licensing Gemini to power the company's A.I. products, though no decisions have been reached. How Meta's A.I. model performs is being closely watched in the competition over the fast-evolving technology. Google, OpenAI and Anthropic are widely regarded as the leaders in frontier A.I. models, which are the basis for developing new chatbots, video generators, coding tools and other products. Being at the forefront of A.I. development also helps companies recruit technologists and keep up a stream of experimentation.

Meta's New AI Image Model Doing Badly in Testing, Report Says

Meta said in January that it plans to spend somewhere between $115 billion and $135 billion in the pursuit of AI superintelligence, but it's hit a snag after its newest AI model is reportedly behind rivals. The New York Times reports that Meta's new AI image and video model, codenamed 'Avocado', has fallen short in tests when compared to models from rivals OpenAI, Google, and Anthropic. The report comes from anonymous sources inside Meta who tell the newspaper that it's failed to beat Gemini 3.0 during internal testing. Avocado was slated to be released this month, but the disappointing results have forced Meta to delay the release of Avocado until at least July. Embarassingly, Meta is now considering licensing Google Gemini to power its AI products in the short term. "As we've said publicly, our next model will be good but, more importantly, show the rapid trajectory we're on, and then we'll steadily push the frontier over the course of the year as we continue to release new models," Meta spokesman Dave Arnold tells The Times. "We're excited for people to see what we've been cooking very soon." Meta has been ploughing resources into AI, investing $14.3 billion in startup Scale AI. CEO Mark Zuckerberg has stated that Meta's new goal is to create AI "superintelligence," where AI will be smarter than humans. The Times reports that, as well as Avocado, Meta is also working on another fruit-themed model called Mango. Both are image and video generators. Meta will likely want its AI models to help out with its booming smart glasses. But that's also hit a snag: the company is now facing a class action lawsuit over the wearable technology, with plaintiffs alleging that the company misled consumers about how footage captured by the devices may be reviewed. The legal action follows a report by Swedish newspapers Svenska Dagbladet and Göteborgs-Posten that contractors tasked with labeling objects in video clips encountered private and sensitive scenes recorded by users. The workers based in Kenya said they had viewed intimate material, including bathroom visits, sexual encounters, and other personal moments.

Meta Is Delaying Its 'Superintelligent' AI Model After Performance Issues

Mark Zuckerberg has invested billions in AI, but Meta's latest model isn't keeping pace with OpenAI and Google. Mark Zuckerberg's $600 billion bet on AI isn't paying off as planned. Meta delayed the release of its new AI model to at least May. The tech fell short of rivals like OpenAI, Google, and Anthropic on internal tests for reasoning, coding, and writing. Meta's model, code-named Avocado, outperformed its previous AI and Google's older Gemini 2.5 model, but couldn't beat Google's latest Gemini 3.0 from November. Zuckerberg has staked Meta's future on leading AI development, declaring last year that the company's new goal was to create a "superintelligent" form of AI that would lead to "a new era for humanity." Meta projected it would spend as much as $135 billion this year on AI -- nearly twice the $72 billion it spent last year. As if losing out to competitors isn't bad enough, Meta's AI division has discussed temporarily licensing Google's Gemini to power the company's AI products while Avocado is delayed. No decisions have been reached.

Meta pushes AI model 'Avocado' rollout to May or later, NYT reports - The Economic Times

Meta has postponed the release of its artificial intelligence model "Avocado" to at least May, from this month, the New York Times reported on Thursday, citing sources. The performance of Meta's new AI model currently falls between Google's AI Gemini 2.5 and ⁠Gemini 3, ⁠delaying its launch until May or June, a person familiar with the matter told Reuters. The delayed timeline comes even as the company invests heavily to expand its AI ambitions, including a roadmap for building its own chips. In January, Meta laid out capital-spending plans of between $115 billion and $135 billion for the year ⁠in the pursuit of "superintelligence" - the horizon where AI will outsmart humans. Meta's new model, which the company has ⁠been working on for months, has fallen short in performance when compared with the latest offerings from rivals, the report said. "Our next model will be good, but more importantly, show the rapid trajectory we're on, and then we'll steadily push the frontier over the course of the year as we continue to release new models," a Meta spokesperson told Reuters. "We're excited for people to see what we've been cooking very soon," the spokesperson added in an emailed statement. The ⁠leaders of Meta's AI division have discussed the possibility of temporarily licensing Gemini to power the company's AI products, the report added, although no decisions have been reached. Media outlets had reported in December that Meta was working on a text AI model code-named Avocado slated for a first-quarter launch.

Meta May Delay 'Avocado' As Tech Underwhelms In Tests Against Google Gemini And Anthropic AI Models: Report - Alphabet (NASDAQ:GOOG), Alphabet (NASDAQ:GOOGL)

Meta Platforms, Inc.'s (NASDAQ:META) upcoming foundational AI model, code-named "Avocado," has reportedly underperformed compared with systems from competitors during internal evaluations. Meta's 'Avocado' AI Model Reportedly Falls Short In Internal Tests The model is designed to power future chatbots, coding assistants and other AI-driven products across Meta's ecosystem. While Avocado performed better than Meta's earlier AI systems and even surpassed an earlier version of Google's Gemini model, it reportedly did not match the capabilities of the latest Gemini release. Launch Delayed As Meta Explores Alternatives Because of the performance gap, Meta has reportedly pushed Avocado's launch back to at least May, from an earlier planned release this month. Executives have also discussed the possibility of temporarily licensing technology from Google's Gemini models to support some of Meta's AI products, though no final decision has been made. Meta did not immediately respond to Benzinga's request for comments. Zuckerberg's Massive AI Bet Zuckerberg has positioned AI as central to Meta's future strategy. The company has spent billions hiring researchers and building infrastructure to support the technology. In January 2026, Meta projected it could spend as much as $135 billion this year, nearly double the roughly $72 billion it spent last year, as it expands data centers and AI capabilities. Meta is also developing additional models, including "Mango," focused on image and video generation. Moreover, a future system is reportedly planned under the code name "Watermelon." Price Action: On Thursday, Shares of Meta fell 2.55% to $638.18 and slipped another 0.18% to $637 in after-hours trading, according to Benzinga Pro. According to Benzinga Edge Stock Rankings, Meta shares are trending lower in the short, medium and long-term, though the company's Quality score were in the 89th percentile. Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors. Photo: PJ McDonnell / Shutterstock Market News and Data brought to you by Benzinga APIs To add Benzinga News as your preferred source on Google, click here.

Meta Pushes Back Avocado AI Launch, Licenses Gemini Tech

Avocado has faced multiple delays over the past few years. Meta insiders initially expected the model before the end of 2025, but it slipped to the first quarter of 2026 amid training and performance-testing challenges, as CNBC first . Bloomberg reported around the same time that Avocado was expected to debut "sometime next spring" and could be launched as a closed, proprietary model -- a departure from Meta's previous open-source Llama strategy. The delay of Avocado AI is the latest setback in Meta's efforts to catch up with other AI competitors. Mark Zuckerberg's AI rivals have long been ahead.

Meta delays Avocado AI model launch amid performance concerns: Report

CEO Mark Zuckerberg has reaffirmed Meta's aggressive AI push, including $115-$135 billion in planned infrastructure investments. Meta has reportedly pushed back the launch of its upcoming AI model, codenamed Avocado. This comes after the New York Times report, citing sources, stated that this model was supposed to launch this month but will now debut in May or later. According to a Reuters report, the delay is linked to performance concerns. The new AI model is currently said to perform between Gemini 2.5 and Gemini 3, developed by Google. However, it has not yet matched the latest capabilities offered by competing AI systems, which has led Meta to reconsider the launch timeline. The report also stated that the company has been working on the Avocado model for several months, but the internal evaluations showed the model still needs improvements before it can compete effectively with rival platforms. Also read: From garage to the world: Tim Cook pens emotional note before Apple turns 50 Even after the delay, the report stated that Meta continues to increase spending on artificial intelligence. Previously, the company announced its plan to invest between $115 billion and $135 billion in capital expenditure as it builds infrastructure and technologies aimed at achieving what it describes as superintelligence. The report also mentioned that a Meta spokesperson also said the company remains confident about the direction of its AI development. The spokesperson reiterated comments previously made by CEO Mark Zuckerberg during the earnings call, stating that the company's next AI model will show big progress that Meta is making in the field. The spokesperson also indicated that additional models are expected to be released over the course of the year. Further, the report also stated that internal discussions at Meta have also explored the possibility of temporarily licensing Google's Gemini technology to support some of its AI products. However, sources familiar with the matter said that no final decision has been made on that front. The Avocado model first surfaced late last year, when media outlets revealed that Meta was developing a new text-focused AI system set to launch in the first quarter. The updated timeline indicates that the company is taking more time to improve the model before making it publicly available.