Meta Scales Ads Recommendation to LLM Level, Doubling Training Efficiency

Meta_Engineers · x · 2026-08-05

Meta officially shared that its Generative Ads Recommendation Model (GEM), powering ads on Instagram and Facebook, is now training at LLM scale across thousands of latest-gen GPUs. Because standard LLM infrastructure doesn't natively fit recommendation workloads, Meta co-designed several custom low-level components:

Through these optimizations, Meta scaled total training FLOPs 4x in 12 months while doubling end-to-end training efficiency to 20–25% MFU.

Related event: Meta Scales Ad Recommendation Models to LLM Size(3 posts)→

Original post →

More from Infra

Infra channel →