Meta is developing a new AI model, code-named Avocado, which may impact the company’s previously established open-source AI strategy, a direction emphasized by CEO Mark Zuckerberg as vital for the future. Reports indicate that access to this new model could potentially come with a fee, marking a shift from earlier practices.
In the prior year, Meta released its Llama 4 AI model, which has faced criticism regarding its open-source designation. The Open Source Initiative has expressed disagreement about whether Llama 4 meets open-source standards. The launch was reportedly underwhelming, overshadowed by allegations that Meta manipulated AI benchmarks and a subsequent delay of a planned “Behemoth” version, as the company shifted focus to new developments.
Zuckerberg, in a memo dated July 30, highlighted the need for a reconsideration of the open-source approach to better manage potential safety risks. He stated that Meta must be cautious about what is made available as open-source technology moving forward.
Additionally, Zuckerberg is reportedly dedicating substantial time to a specialized group within the company known as TBD Lab. This new team is situated near his office at Meta’s headquarters and is tasked with advancing the company’s AI initiatives. In light of these developments, it remains to be seen how Meta will balance its open-source commitments with the financial imperatives tied to its evolving AI strategy.
Source: https://www.theverge.com/news/841809/meta-avocado-ai-model-charge-open-source-mark-zuckerberg