engblogs

summaries of the latest blog articles from your favorite tech companies.
MetaMeta

Zoomer: Powering AI Performance at Meta’s Scale Through Intelligent Debugging and Optimization

Zoomer is Meta's automated platform for AI performance profiling, debugging, and optimization across large-scale GPU infrastructures, driving significant efficiency, energy savings, and throughput improvements in training and inference workloads.

11/21/2025
Lambda LabsLambda Labs

Silicon Photonics for AI Clusters: Performance, Scale, Reliability, and Efficiency

Exploring how silicon photonics and co-packaged optics revolutionize AI GPU clusters by enhancing performance, scalability, reliability, and energy efficiency in large-scale compute networks.

11/21/2025
OpenAIOpenAI

Early experiments in accelerating science with GPT-5

GPT-5 accelerates scientific discovery by enabling expert collaboration across disciplines, delivering rapid insights from mathematics, biology, and physics, and highlighting both the capabilities and limitations of AI-assisted research.

11/20/2025
OpenAIOpenAI

Early experiments in accelerating science with GPT-5

GPT-5 is accelerating scientific discovery by assisting expert researchers with literature review, proof generation, hypothesis formulation, and experimental design across diverse fields, while recognizing the need for human oversight and collaboration.

11/20/2025
OpenAIOpenAI

Helping 1,000 small businesses build with AI

OpenAI's Small Business AI Jam empowers over 1,000 small business owners across five U.S. cities with hands-on AI training and tailored tools to enhance growth, productivity, and competitiveness.

11/20/2025
OpenAIOpenAI

OpenAI and Foxconn collaborate to strengthen U.S. manufacturing across the AI supply chain

OpenAI and Foxconn join forces to co-design and manufacture advanced AI infrastructure hardware in the U.S., aiming to strengthen domestic supply chains and accelerate AI innovation.

11/20/2025
SalesforceSalesforce

Solving Real-Time AI Classification for Agentforce: How Single-Token Prediction Delivers 30x Faster Agent Responses

HyperClassifier leverages single-token prediction and KV caching to achieve 30x faster real-time AI classification for Agentforce, balancing speed and accuracy without relying on human-labeled data.

11/20/2025
MetaMeta

Key Transparency Comes to Messenger

Messenger introduces key transparency verification to enhance the security and authenticity of end-to-end encrypted chats by allowing automatic and auditable confirmation of public encryption keys.

11/20/2025
Modular AIModular AI

Modular: Modular 25.7: Faster Inference, Safer GPU Programming, and a More Unified Developer Experience

Modular Platform 25.7 delivers faster AI inference, enhanced safety in GPU programming with Mojo, expanded hardware support including NVIDIA Grace superchips and Apple silicon GPUs, and a more unified developer experience through open-source MAX APIs and advanced experimental model workflows.

11/20/2025
Google DeepMindGoogle DeepMind

Introducing Nano Banana Pro

Introducing the innovative Nano Banana Pro, a cutting-edge solution designed to enhance your technical projects with unparalleled efficiency.

11/20/2025
Google DeepMindGoogle DeepMind

Build with Nano Banana Pro, our Gemini 3 Pro Image model

Explore how to build projects using the Nano Banana Pro with the Gemini 3 Pro Image model for enhanced performance.

11/20/2025
Google DeepMindGoogle DeepMind

How we’re bringing AI image verification to the Gemini app

Introducing AI-powered image verification to enhance security and reliability in the Gemini app.

11/20/2025