Descriptions:
Internal memos obtained by The Information reveal that Meta has completed pre-training on a new flagship model codenamed Avocado, developed under the company’s restructured Meta Superintelligence Labs. This TheAIGRID video breaks down what those memos disclose and why the timing matters.
According to a January 20th memo signed by Superintelligence Labs product manager Megan Fu, Avocado is already performing competitively with leading post-trained models on knowledge, visual perception, and multilingual benchmarks — before any post-training refinement has been applied. A separate December memo claims Meta achieved 10x compute efficiency gains on text tasks compared to its Maverick model and over 100x efficiency improvements versus Behemoth, the large Llama 4 variant Meta quietly shelved last year.
The video contextualizes these claims against Meta’s credibility problem: in 2025, Llama 4 was widely criticized for manipulated benchmark results, Yann LeCun acknowledged the figures were inflated, and Mark Zuckerberg ultimately dismantled the entire generative AI team to build the Superintelligence Labs from scratch. That history raises the stakes for Avocado’s eventual public release — another benchmark controversy, the video argues, could be terminal for Meta’s standing in the frontier model race. Whether the internal optimism survives post-training and public evaluation remains to be seen.
📺 Source: TheAIGRID · Published February 08, 2026
🏷️ Format: News Analysis
