Meta's AI lab pushes ahead with Llama 4.X release
Meta's Superintelligence Lab, with an internal team known as TBD, is gearing up to release its first big project, Llama 4.X, by the end of 2025.
This new AI model aims to fix the problems that held back earlier versions, which got mixed reviews for underperforming.
Even after some top team members left and previous models Scout and Maverick (launched in April) didn't wow everyone, the lab is doubling down on upgrades and bug fixes.
Competing with the best in AI
The goal? Create superintelligent AI that can compete with the best from OpenAI and Google DeepMind.
Since launching in June, the lab has attracted talent from those rivals and worked on Llama 4.1, 4.2, and next-gen tech.
Mark Zuckerberg confirmed steady progress last July.
Despite bumps along the way, Meta's team is determined to make Llama 4.X ready for real-world use by the end of this year (2025)—showing they're serious about moving AI forward.