Meta is preparing to radically change its strategy in the field of artificial intelligence. According to Bloomberg, the company plans to abandon the open-source policy it has supported for years and switch to a closed and paid system similar to competitors like Google and OpenAI. The new model, codenamed Avocado, is expected to be introduced in the spring of 2026. Unlike previous releases, this system will not be available to everyone; instead, access will be strictly controlled and offered through a paid subscription.
The main reason behind this sharp shift is the company’s previous failure. The Llama 4 model, introduced in early 2025, disappointed both Silicon Valley and Mark Zuckerberg. After the project failed to meet expectations, Zuckerberg restructured the team and redirected the focus toward projects capable of generating direct revenue. He is currently working with a new team called TBD Lab and is personally overseeing the process.
Believing that the company has fallen behind in the AI race, Meta’s leadership has launched an aggressive hiring strategy to turn things around. The company has offered tens of millions of dollars in salary to recruit the head of Apple’s AI division, as well as experienced researchers from leading companies such as OpenAI and Anthropic. In addition, former GitHub CEO Nat Friedman will oversee product development, while Scale AI’s founder Alexandr Wang will lead the superintelligence lab.
If you want, I can also create a shorter summary, a headline, or a more formal news-style version.
