Meta AI

GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model

Meta’s Generative Ads Recommendation Model (GEM), the foundation model behind ads recommendations across Instagram and Facebook, now trains at LLM scale on several thousand of the latest-generation GPUs.

Meta AI||1 min read
Open original

At a glance

Source
Meta AI
Published
Aug 3, 2026
Read time
1 min read
Primary lane
Ai Research

Quick read

3 bullets
  • Meta’s Generative Ads Recommendation Model (GEM), the foundation model behind ads recommendations across Instagram and Facebook, now trains at LLM scale on several thousand of the latest-generation GPUs.
  • This post goes into the details on how we achieved: doubling end-to-end (E2E) training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x in [...] Read More...
  • The post GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model appeared first on Engineering at Meta .

Why it matters

Product updates matter when they expand what builders and teams can actually ship now. A release is important if it changes capability, usability, or deployment economics in a way that affects real workflows.

Builder takeaway

Meta AI published this update in the Ai Research lane. Use the original source for details, then compare it with related briefings before changing a roadmap, workflow, or production system.

Product updates matter when they expand what builders and teams can actually ship now. A release is important if it changes capability, usability, or deployment economics in a way that affects real workflows.

Stay ahead with daily AI briefings

Follow the feed, share the briefing, or jump back into the archive.