Meta has postponed the release of its artificial intelligence model Avocado to at least May, from this month, according to a report by the New York Times. A person familiar with the matter told Reuters that the performance of the new model lies between Google’s Gemini 2.5 and Gemini 3.
The NYT report mentioned that 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.
“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,” a Meta spokesperson told Reuters, reiterating CEO Mark Zuckerberg’s remarks during earnings call in January. “We’re excited for people to see what we’ve been cooking very soon,” the spokesperson added.
READ: Inside Moltbook: The social platform Meta just acquired (
The report also mentioned that the leaders of Meta’s AI division have discussed the possibility of temporarily licensing Gemini to power the company’s AI products. However, the report added, although no decisions have been reached.
The development of the AI model has been reportedly closely watched by investors. Meta Chief Executive Mark Zuckerberg spent billions last year recruiting top engineers to shake-up the leadership of his AI development labs.
The move came after Meta’s previous Llama model faced delays and was also outperformed by rival AI models.
Zuckerberg had said on a call with analysts in late January that he expects the company’s first Avocado model will be “good but, more importantly, show the rapid trajectory we’re on.”
READ: Meta: Parental oversight not enough to reduce teen social media compulsion (
“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 told the New York Times. “We’re excited for people to see what we’ve been cooking very soon.”
The delayed timeline comes as Meta invests heavily to expand its AI ambitions, including a roadmap for building its own chips. In January, Meta had 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.
That money is focused on building data centers to train and run AI algorithms. Investors turned negative on such spending late last year, questioning whether Meta could see strong enough returns at its current spending levels.