Meta Delays Launch of Avocado AI Model, May Consider Licensing Google’s Gemini

Meta Delays Avocado AI Model, Considers Licensing Google’s Gemini

Meta has delayed the release of its next-generation artificial intelligence model, known internally as Avocado, pushing its launch from March to at least May 2026. The delay comes as the company works to improve the model’s performance, which currently falls short of competing systems.

According to reports citing sources familiar with the matter, Avocado’s performance is currently positioned between Google’s Gemini 2.5 and Gemini 3. This places it below the level Meta had hoped to achieve in its effort to compete with leading AI developers.

The delay is notable because Meta had previously presented early 2026 as a key moment to strengthen its position in the global AI race. The company has invested heavily in AI development, with plans to spend between $115 billion and $135 billion on AI infrastructure and research in 2026.

In response to the delay, Meta is reportedly considering temporarily licensing Google’s Gemini model to power some of its AI products while Avocado continues development. If implemented, the move would allow Meta to maintain competitive AI capabilities across its platforms, including Facebook, Instagram, and WhatsApp.

The possibility of using a rival company’s technology is significant, given Meta’s long-standing focus on building its own AI systems. The company previously promoted its Llama family of open-source models, which gained popularity among developers and helped establish Meta’s presence in the AI ecosystem.

Avocado is being developed by Meta’s TBD Lab, a new research division formed to accelerate the company’s AI progress. The team includes several high-profile hires, such as former GitHub CEO Nat Friedman and AI researcher Shengjia Zhao, one of the co-creators of ChatGPT.

However, the project has already experienced multiple delays. Originally planned for release in 2025, the launch was later moved to early 2026 before being pushed back again.

Despite the setbacks, Meta has emphasized that development is still progressing. A company spokesperson said the upcoming model will demonstrate the “rapid trajectory” of Meta’s AI capabilities.

The delay comes amid growing competition in the AI industry. OpenAI is developing GPT-5, while Google’s Gemini 3 and Anthropic’s Claude 4 continue to advance, increasing pressure on Meta to deliver a competitive AI model in the coming months.