Meta Reportedly Delays “Avocado” AI Model Release, Now Tipped for May or June

In the ongoing AI race, Meta might be feeling a little left out, given the failure of Llama models and the fact and that it hasn’t pushed new models for a long time now. On the other hand, its rivals, like Google, Anthropic, and OpenAI, are pushing one model after another. However, Meta hopes that 2026 will change its fate, as it reportedly has a few models coming this year.

The release delay has been linked to performance concerns

The upcoming AI model from Meta, code-named Avocado, which was previously scheduled to launch in March, has now reportedly been delayed to May or June, according to The New York Times (via Reuters). Per the NYT, the delayed release has been linked to performance concerns, as the model currently sits somewhere between Google’s Gemini 2.5 and Gemini 3 in capability.

As a company making a comeback, Meta cannot afford to launch an underperforming model. Do note that the delay in the release of the Avocado model comes despite Meta’s aggressive AI investment strategy. Earlier this year, the company outlined plans to spend $115 billion to $135 billion in 2026, to accelerate progress toward what CEO Mark Zuckerberg describes as “superintelligence.”

Meta is also exploring long-term infrastructure improvements, including developing its own AI chips to reduce reliance on external hardware.

During earning call earlier this year, a Meta spokesperson told Reuters that the upcoming model will demonstrate the “rapid trajectory” of the company’s AI development and hinted that more models could follow throughout the year.

Speaking of Meta, let’s not forget that the company has also started rolling out anti-scam tools across its social media apps, including Facebook, WhatsApp, and Messenger.