Meta is ditching its long-standing open-source approach to artificial intelligence and is now building a paid AI model dubbed ‘Avocado,’ a move that marks a major turnaround in how it makes money from AI (via Bloomberg).
According to people familiar with the matter, Meta’s proprietary new AI model is due to be released next spring and will likely offer restricted access that users must pay for. This change underlines a shift from distributing free AI tools to pursuing a business model more akin to what rivals such as OpenAI and Google employ.
Since launching the open-source version of its Llama models, Meta had positioned itself as a champion of accessible AI. But slower-than-expected adoption and internal dissatisfaction, particularly around the release of Llama 4, appear to have pushed executives to reconsider that strategy.
Meta’s CEO Mark Zuckerberg has taken a hands-on role in the company’s new AI push, personally engaging with top engineers at a newly formed team known as “TBD Lab.” The company has reportedly invested heavily in talent, including hiring from leading AI outfits and offering multiyear compensation deals worth hundreds of millions, reflecting how aggressively it is betting on proprietary AI.
Under the new direction, Meta has quietly stashed away several open-source AI efforts. Some employees have been advised to avoid public discussion of open-source projects like Llama, and certain initiatives have been paused entirely. That has included shelving a powerful internal model, previously known as Behemoth, purportedly because its performance did not meet expectations.
Avocado is said to draw from a mix of open-source and third-party models during training, including systems developed by Google, OpenAI and a Chinese firm under Alibaba. Meta’s new Chief AI Officer Alexandr Wang, who arrived after Meta’s $14.3 billion investment in his former company, is believed to lead its development.
Meta’s change in course signals a recognition that continuing with free models may no longer align with its broader business ambitions, especially given the steep cost of data centers, infrastructure and high-end talent.
Want to see more of our stories on Google?
P.S. Want to keep this site truly independent? Support us by buying us a beer, treating us to a coffee, or shopping through Amazon here. Links in this post are affiliate links, so we earn a tiny commission at no charge to you. Thanks for supporting independent Canadian media!
