Muse Spark 1.3 is now available through @Oracle, with availability on @GoogleCloud in private preview. We’re excited to give developers more choice in where they build and deploy by expanding access to our frontier coding and agentic model through our preferred cloud platform partners.

Last year we introduced Private Processing, Meta’s confidential computing infrastructure that keeps everything you share with your AI assistant private, so that no one, not even Meta, can access your data. At #MetaConnect, we announced that we’re bringing these same privacy benefits to our AI glasses. Read the technical deep dive on our approach: engineering.fb.com/2026/09/2…

Today we’re announcing Petal: the world’s first petabit-class transoceanic subsea cable and the first to deploy multi-core fiber at scale. Spanning 7,000 km (4,000+ mi) between France and the United States, Petal will deliver 1 Petabit per second (1 Pbps / 1,000 Tbps), doubling the capacity of today’s most advanced transatlantic systems. Petal is set to enter service in 2029, setting a new standard for high capacity, high density SDM subsea cable designs. Read the full technical breakdown on our engineering blog: engineering.fb.com/2026/09/2…

To help Meta subject matter experts focus more on work where their knowledge matters most, we’ve built an AI agent that acts as a secondary expert for a given domain. This system makes deep specialist knowledge readily available and preserved for anyone in an organization to access, share, and build upon. Two things make it novel and portable: 1️⃣ structured, auditable knowledge architecture that separates what the agent knows from how it reasons 2️⃣ self-improvement loop that turns expert feedback into verified updates without retraining We’re sharing a deep dive into our approach and architecture here: engineering.fb.com/2026/09/0…

How do you manage database connections when you have 1M+ clients hitting 500k+ database hosts? We’re sharing a deep dive into ZGateway, the stateless proxy tier handling traffic for ZippyDB (>1B ops/sec). The direct-access model created an unbounded TLS connection mesh, wasting CPU and causing OOM crashes during reconnection storms. ZGateway decoupled the fleets, dropping per-host connections by 97–98% with just a 6% compute overhead. Read the full engineering deep dive here: engineering.fb.com/2026/09/0…

Introducing Muse, your personal AI agent from Meta that gets things done across every part of life. Download the Muse app and get started: Muse.ai

Introducing Muse, the personal agent that understands your goals and works 24/7 to get things done for you.

Training and serving frontier AI models depends on fast, reliable networks that move data between GPUs without wasting compute cycles. To meet this challenge at scale, we designed MetaRoCE: a clean-sheet RDMA transport protocol purpose-built for AI workloads on commodity Ethernet. We’re releasing the MetaRoCE specification, a reference software implementation and a compliance test suite through the @OpenComputePrj to enable the broader industry to adopt, implement, and build on it. Read the full technical deep dive: engineering.fb.com/2026/08/2…

The convergence of sequence modeling and recommendation systems has opened new frontiers in artificial intelligence. 🚀 Our new multi-layer sequence learning architecture demonstrates LLM-like scaling law, while delivering performance in latency-sensitive ads ranking models. This new architecture is deployed for user representation within Meta’s ads recommendation as a resource efficient, multi-stage system that boasts high upstream-to-online transfer ratios. 👉 Read our research paper to learn more: engineering.fb.com/2026/08/0…

Meta’s Generative Ads Recommendation Model (GEM), the engine behind ads on Instagram and Facebook, is now training at LLM scale across thousands of the latest-gen GPUs. Because standard LLM infrastructure doesn't natively fit recommendation workloads, we co-designed custom kernels like Jagged Flash Attention (JFA), Generalized Dot-Product Attention (GDPA), BlockAttention, ultra-low mixed precision (including MXFP8 attention), and a topology-aware 5D parallelism architecture tailored to our network hierarchy. The result: we scaled total training FLOPs 4x in 12 months while doubling end-to-end training efficiency to 20–25% MFU. Read the full deep dive here: engineering.fb.com/2026/08/0…

When a Linux kernel upgrade risked regressing our ads fleet, we turned to sched_ext — an open-source scheduler that ships updates in days, not months. The result: 28% lower tail latency, 3.28 MW saved, and a 1.1% increase in the number of ads ranked. Now it's upstream in Linux v6.12 for everyone. Read the technical blog post: engineering.fb.com/2026/07/1…

AI compute performance has roughly tripled every two years, but storage growth remains modest, making I/O bottlenecks a primary cause of GPU stalls. To maximize GPU utilization and research velocity across hundreds of exabyte-scale storage clusters, we overhauled Meta’s BLOB-storage architecture specifically for modern AI workloads. Read a full technical deep dive: engineering.fb.com/2026/07/0…

Facebook Groups hold massive amounts of community knowledge, but finding it depended on typing the exact right keywords. Not anymore. We've modernized Facebook Groups Search with a new hybrid retrieval architecture, yielding measurable improvements in search engagement and relevance without increasing error rates. Learn about the technical approach: engineering.fb.com/2026/04/2…

This year marks Meta’s 10th consecutive year as a sponsor of @ThePSF! Python remains the most used programming language across our engineering stack, powering the backends of Instagram and Threads alongside our AI research infrastructure. By supporting the PSF, we aim to help ensure that Python remains robust, innovative and accessible for generations of engineers to come. Read our full reflection on our long-term commitment to Python: engineering.fb.com/2026/06/3…

How do you fit all-day battery life, cameras, speakers, and AI workloads into a pair of sleek smart glasses? You completely rethink how batteries are built. In the latest Meta Tech Podcast, Meta engineers Karthik and Myuran reveal how they designed custom 7mm steel-can batteries – narrower than an adult’s pinky finger – to power the Ray-Ban Meta and Oakley Meta Vanguards. Tune in to discover: - How they eliminated "brownouts" during heavy AI multitasking - The systems puzzle of balancing two separate batteries in a single pair of glasses - Driving 2x the runtime out of a 30% cell capacity bump 🎧 Listen to the full episode now: engineering.fb.com/2026/06/2…

Most LLM classification failures aren't prompt failures, they're context failures. Passing dozens of raw fields (like a column named "age") causes high token usage and diluted model attention. We solved this at Meta by building a "context mesh" that ranks signals into structured evidence briefs before prompting. Dive into the 7-stage operational pattern we use to classify complex, AI-native data modalities: engineering.fb.com/2026/06/2…