Intel Devs

🚀 optimum-intel 2.0 is out! The @intel optimization toolkit for 🤗 Transformers just went OpenVINO-first ⚡ pip install optimum-intel → OpenVINO + NNCF built in, no extras 🤝 Transformers v5 ready 🧠 Gemma 4, Qwen3.5/3.6, Qwen3-VL, Arcee Trinity & more 🗜️ Data-aware AWQ, default INT8, Eagle3 speculative decoding Run the newest open models fast on Intel CPUs, Arc GPUs & NPUs 👇 github.com/huggingface/op… Congrats @ellacharlaix and team @OpenVinoAI @intel Blog post in 🧵

From live video to real-time insight. DL Streamer ingests camera streams, normalizes frames for multimodal AI, runs inference, and outputs via RTSP/MQTT—powering real-time monitoring at the edge. Learn more: ms.spr.ly/6011vgC3T

Object detection at scale requires more than just a model. Our second OpenVINO DevCon workshop walks through annotation, training, optimization, and deployment for real-world manufacturing. Join the session to build your skills and earn a Credly badge. Register at ms.spr.ly/6015vgLad

Set up a local LLM for free in VS Code. Learn how to deploy a Qwen3-Coder model with OpenVINO Model Server and connect it to the Continue extension for chat, autocomplete, and agentic workflows—all inside your IDE.

Prototyping object detection is fast, but scaling it into a production pipeline... not so much. This OpenVINO DevCon workshop covers tuning, optimization, and deployment on Intel Core and Intel Core Ultra systems. Save your seat: ms.spr.ly/6014v5cF6

What changes when video generation runs on real hardware? Temporal consistency, longer pipelines, and heavy memory demands. See how this becomes controlled, repeatable runs on local systems. Watch the OpenVINO 2026 workshop on demand at ms.spr.ly/6011v5R2H

At COMPUTEX 2026: a few of our partners’ industry-ready robot innovations—driven by Intel dev tools along with pre-validated hardware reference boards, sensor partner enablement, and dev kit ecosystems. See how to put robotics into production faster: ms.spr.ly/6018vdeMM

Heading to CVPR? Come hang with us Friday night! Food, drinks + live AI demos with OpenVINO, Ultralytics, and OpenCV, and it’s right across from the convention center in Denver. It’s a relaxed evening with folks building interesting things in computer vision and AI. Free to attend! Come say hi. ms.spr.ly/6012vjsuC

Take your ecosystem from fragmentation to open adoption. At COMPUTEX 2026, see how Intel Robotics helps organizations like @AAEON build faster and more consistently with dev kits, pre-validated boards, and more. Learn more: ms.spr.ly/6016vd3VI

At CVPR 2026, "From Perception to Action: Building Efficient and Deployable Robot Intelligence Pipelines with Open-Source Edge AI Toolkits" combines 3D vision, representation learning, and policy learning to build a real-time table manipulation system on low-cost arms at the edge—while demonstrating how teams can replicate the process with open-source edge AI toolkits.

At @RedHat Summit 2026, Team Intel brought enterprise AI to life with AI-powered robots and real-world edge AI demos across the Intel and Red Hat ecosystems. See how teams can modernize and scale AI: ms.spr.ly/6012vbptQ

The future of physical AI is on display at COMPUTEX 2026. OpenVINO 2026.2 is available now and features support for Hugging Face Transformers v5, custom ops, GPU KV cache compression, and new physical AI capabilities. Download today: ms.spr.ly/6016vbpaE

Live from COMPUTEX 2026: Catch @TrossenRobotics' bimanual robot, built with OpenVINO Physical AI and #IntelCoreUltra Series 3 to manipulate assembly parts, see and act to configure those elements, and demonstrate cutting-edge physical AI robotics capabilities. This combination of Intel technology enables vision-language-action AI to run entirely at the edge with low-latency and responsive decision-making for physical tasks. Try OpenVINO Physical AI, available now: ms.spr.ly/6010vb9ly

Robotics innovation runs on Intel. It’s been true since the early days of industrial robots. And it will be true tomorrow, as Intel drives open robotics and accelerates physical AI in real-world deployments across industries. See it in action at COMPUTEX 2026. Learn more: ms.spr.ly/6012vbpvq

@bobduffy Great walkthrough, Bob! It's amazing to see everything you can do remotely with AI Playground Home Agent. It makes your phone an even more powerful connecting tool.

🚨Video Walkthrough of AI Playground Home Agent! Take the full power of your home AI PC anywhere in the world! 🌍💻 In this video, Bob Duffy walks through the early access and setup of AI Playground Home Agent from @Intel, being showcased live at Computex 2026! 🤖 What is AI Playground Home Agent? Home Agent is a brand-new experimental feature that allows you to remotely prompt your local AI Playground installation from anywhere. Using a secure Telegram chat on your phone, Home Agent bridges the gap between your mobile device and your desktop, letting you leverage the massive, cost-free AI compute power of your own hardware while you’re on vacation, traveling, or just relaxing on the couch. 🚀 Key Features You Can Control Remotely: Advanced Chat & Vision: Have deep conversations and analyze images on the go. - Document Search: Access and query your local files securely from your phone. - Image Gen & Magic Editing: Create and modify visuals using your desktop's GPU. - Video Generation: Render video clips remotely without draining your phone's battery. - MCP Application Control: Use the Model Context Protocol (MCP) to control local applications running on your home PC. 🖥️ Hardware Supported: Get maximum performance out of your local hardware: -Intel® Core™ Ultra processors -Intel® Arc™ GPUs -NVIDIA GeForce RTX™ GPUs 📥 Get Early Access: Home Agent is available starting in AI Playground v3.1.1-beta or newer. 👉 Download it here: intel.com/ai-playground #Intel #AIPlayground #Computex2026 #CoreUltra #IntelArc #LocalAI #GenerativeAI #RTX

Reduce the complexity of deploying AI models and robot policies with a single, unified open-source API. OpenVINO Physical AI, part of OpenVINO 2026.2, helps developers standardize and deploy at scale. Among other advancements, it features: • A common interface that supports multiple physical AI model architectures • Hardware-accelerated OpenVINO inference for physical AI • Asynchronous and RTC (real-time chunking) inference support • Security‑ and safety‑first architecture with future‑ready hooks Learn more: ms.spr.ly/6014vZruI

Intel has shaped robotics for decades. Now, that vision comes together under a new brand: Intel Robotics. It’s bigger than just a name—it’s an open, scalable robotics enablement strategy designed for the age of robotics and physical AI. And it’s a commitment to empower developers to build faster, more seamlessly, and smarter than ever before with: • Better performance • Stronger precision motion control (determinism) • Easier integration across Intel architectures Learn more: ms.spr.ly/6012vkt5a

Throughout 2026, DevCon will present a series of eight webinars to help developers build, optimize, and deploy real AI applications. Watch session 1 on demand now for a summary of the OpenVINO roadmap for 2026, insights into key AI trends, and an exploration of how teams can optimize workloads ranging from LLMs like gpt-oss and Qwen3-30B to video generation and real-time computer vision with YOLO26. Register today at ms.spr.ly/6018vZF7a

Train vision AI to understand what it’s never seen. Geti Instant Learn combines semantic understanding with simple prompts for zero-shot learning, plus deterministic edge-optimized execution for real-world adaptability. See how: ms.spr.ly/6013vZ3wT

Intel’s commitment to open robotics runs deep, and it's helping developers improve performance and bridge the gap between prototype and production. At the system and software layer of the robotics stack, Intel is contributing upstream optimizations to ROS 2—for better performance, real-time responsiveness, and precision motor control on Intel silicon. At the same time, Intel’s Robotics AI Suite helps ease the process of production scaling by integrating ROS 2 with SLAM, vision-language-action models, and modern AI and by validating on Intel platforms. Learn more: ms.spr.ly/6019vZyev