Stay connected with BizTech Community—follow us on Instagram and Facebook for the latest news and reviews delivered straight to you.
Meta has postponed the launch of its next-generation artificial intelligence model, internally code-named Avocado, after internal evaluations indicated it failed to match the performance of competing systems such as Gemini.
The model, originally expected to debut this month, has now been delayed until at least May as engineers work to improve its capabilities. The decision represents a setback for Mark Zuckerberg, who had pledged in mid-2025 that Meta’s upcoming AI systems would “push the frontier” of artificial intelligence within the following year.
Sources close to the project said early testing suggested Avocado outperformed Meta’s previous systems and even surpassed earlier versions of Google’s Gemini models. However, once benchmarked against Gemini 3.0, released in November 2025, the system fell short in key performance metrics.
The shortfall prompted internal discussions about whether Meta should temporarily license Google’s Gemini technology to support certain AI-powered features while Avocado undergoes further development. No final decision has been taken, and Meta has declined to comment directly on the reported timeline change.
A company spokesperson reiterated that Meta remains committed to building “the most advanced and useful AI systems,” but did not address details of the delay.
Rising Pressure in the AI Race
The postponement underscores the intense competitive pressure among technology companies racing to dominate the rapidly evolving AI sector. Alongside Google, firms such as OpenAI and Anthropic have released increasingly powerful models in recent months, raising the performance bar across the industry.
Meta has invested heavily in AI infrastructure to remain competitive. The company has committed billions of dollars toward building new data centres and developing custom semiconductor chips designed to train and run large-scale AI systems more efficiently.
It’s widely known that Llama models, which are released under an open-source framework, have attracted significant developer interest. Avocado, however, was expected to represent a more ambitious leap forward, particularly in reasoning, multimodal understanding and advanced generative capabilities.
Industry analysts say that delays in frontier AI projects are increasingly common given the complexity of training models with hundreds of billions — or even trillions — of parameters. Even minor gaps in benchmark performance can justify postponing a release, particularly when competing systems advance rapidly.
Implications for Meta’s Platforms
The delay may temporarily slow Meta’s plans to integrate more sophisticated AI capabilities across its ecosystem of consumer platforms, including Instagram, WhatsApp, and Facebook.
Artificial intelligence has become central to Meta’s strategy for boosting user engagement, improving content recommendations and expanding new services such as AI assistants and automated content generation.
Competitors have already begun embedding advanced AI models into their products. Google has incorporated Gemini 3.0 features across its search and productivity tools, while OpenAI continues refining its GPT-series systems for both consumer and enterprise applications.
Market Reaction and Industry Outlook
Financial markets reacted calmly to the news, with Meta’s shares largely unchanged in early trading. The company’s valuation remains above $1.5 trillion, reflecting investor confidence in its long-term position within the AI economy despite short-term development hurdles.
Nevertheless, the delay highlights the extraordinary costs and technical challenges associated with building cutting-edge AI models. Training such systems requires immense computing resources, specialised hardware and vast datasets — factors that make the field increasingly dominated by a handful of well-capitalised technology giants.
For Meta, the Avocado setback is unlikely to derail its broader AI ambitions, but it does illustrate the unpredictable nature of technological breakthroughs at the frontier of machine learning.
As the global AI race accelerates, even the largest companies must navigate a reality in which progress is uneven, expectations are high and competitive advantages can shift quickly. Meta’s next few months of development will therefore be closely watched across the technology industry as it seeks to regain momentum in one of the most consequential technological contests of the decade.
Read Also: China Records Sharp Rise in Number of Generative AI Users
