Inside Meta’s AI struggle: Why much-hyped model ‘Avocado’ is facing delays

Meta’s highly anticipated frontier AI model, Avocado, has hit a major setback in its effort to deliver a model that can truly compete with the market’s current leaders. According to a report by the New York Times, the company has officially pushed back the release of "Avocado" from this month to at least May. Currently, the model’s performance falls between Google’s AI Gemini 2.5 and Gemini 3, leading the Meta to delay the launch ahead by at least two months to further refine its capabilities.

The recent revelation comes as the company invests heavily to expand its AI capabilities, including a roadmap for generating its own chips.

Earlier this year, Meta laid out capital-spending plans of between $115 billion and $135 billion for the year in search of “superintelligence”-the horizon where AI will outpace human intelligence. Meta has been working for months to deploy the new model, but it has fallen short in performance when compared with the latest offerings from rivals.

In this regard, a Meta spokesperson told Reuters, reiterating CEO Mark Zuckerberg’s remarks during an earning call in January: “As we’ve rapidly said publicly, our next model will be good, but more importantly, show the rapid trajectory we’re on, and then we’ll as steadily push the frontier over the course of the year as we continue to release new models.”

The spokesperson further clarified the situation, stating: "Despite the delay, we’re really excited for people to see what we've been cooking very soon.” Prominent leaders within Meta’s AI division have shed light on the delay and discussed the possibility of temporarily licensing Gemini to power the company’s AI products in the interim. 

However, no final agreement has been reached regarding the launch timeline. Several media outlets have been reporting that Meta was working on a text AI model, code-named Avocado, which was originally scheduled for a first-quarter release.