Meta has reportedly pushed back the launch of its upcoming text-based large language model, internally known as Avocado, with the release now expected no earlier than May, according to recent reports.
The model had originally been planned for rollout this month. However, sources indicate the system has not yet reached the performance levels Meta had hoped for, particularly when compared with competing AI models such as Google’s Gemini 2.5 and the newer Gemini 3.
The delay comes even as Meta continues to pour massive resources into artificial intelligence development. Earlier this year, the company revealed plans to invest between $115 billion and $135 billion in capital spending, part of its broader effort to push toward what it describes as “superintelligence,” a stage where AI systems could exceed human capabilities.
A Meta spokesperson said the company’s upcoming model will demonstrate meaningful progress, but the bigger focus is the pace at which its AI technology is advancing. The representative noted that Meta plans to keep pushing the limits of its models throughout the year by steadily releasing new systems and upgrades.
The spokesperson also hinted that the company is preparing to reveal new developments soon, saying the team is eager for people to see what it has been building.
Meanwhile, internal conversations within Meta’s AI group have reportedly explored the idea of temporarily licensing Google’s Gemini technology to power certain Meta products. No final decision has been made regarding that possibility.
Reports late last year had already indicated that Meta was developing Avocado as a text-focused AI system, with an initial target launch in the first quarter.
In addition to Avocado, Meta is also said to be working on another AI project internally called Mango, which is designed to specialize in image and video generation. The model is believed to be a key component of the company’s expanding generative AI strategy.
Also Read: Meta Launches Applied AI Engineering Team to Advance Superintelligence Efforts
