NVIDIA AI Developer

If you’re seeing this, you should be following @NVIDIAAI. Our developer team has moved next door to @NVIDIAAI. Join us there to stay up to date on the latest products, models, deep dives, and more. Thanks to everyone who’s been with us here. We hope to see soon 💚

Reminder ❗️ Our developer team has moved. For the latest updates be sure to follow @NVIDIAAI. See you there!

Dev Community Live: SJSU Hackathon Winners – Building Impactful Agents x.com/i/broadcasts/1…

Scaling NemoClaw: Roadmap, OpenClaw Collaboration, and Real‑World Integration | Nemotron Labs x.com/i/broadcasts/1…

Meet Nemotron 3 Nano Omni 👋 Our latest addition to the Nemotron family is the highest efficiency, open multimodal model with leading accuracy. 30B parameters. 256K context length. 🧵👇

📊 Day 0 performance is here: DeepSeek-V4-Pro running on NVIDIA Blackwell Ultra. Using @vllm_project's Day 0 recipe, we’ve captured the initial performance Pareto for DeepSeek’s flagship 1M long-context model. This curve highlights the baseline for balancing AI factory throughput with real-time user interactivity. This is just the starting line. Expect these numbers to climb as we optimize the full co-design stack, including: • NVFP4 & Dynamo • Optimized CUDA kernels • Advanced parallelization techniques and beyond Read the full technical deep dive: nvda.ws/4u0gCcc

Happy Friday! We just put DeepSeek-V4-Pro up on build.nvidia.com. It’s the world’s largest open source model at 1.6T parameters, and you can run it for free running on NVIDIA Blackwell GPUs. Try the NVIDIA NIM API → build.nvidia.com/deepseek-ai/de…

✨ DeepSeek-V4 is here — a million-token context, 1.6T parameter powerhouse optimized for agentic workflows. Out of the box, on DeepSeek-V4-Pro, NVIDIA Blackwell Ultra delivers over 150 TPS/user interactivity for agentic workflows. And we’re just getting started. Expect these performance figures to climb higher as we implement Dynamo, NVFP4, and advanced parallelization techniques. Start building today with @lmsysorg and @vllm_project

DGX Spark Live: Ask the Experts - Gemma 4 on DGX Spark x.com/i/broadcasts/1…

Nemotron Hackathon Winners - S Korea x.com/i/broadcasts/1…

What’s possible with Gemma 4 + DGX Spark? Tune in Friday as NVIDIA and Google DeepMind experts walk through vision translation, long-context document Q&A, and real-time code gen demos live. Bring your questions. 📆 Friday, April 24 at 11:00 AM PDT Add to schedule ➡️ nvda.ws/4cuvGcl

I have had early access to GPT 5.5 for over a week now. And have been cooking with @NaderLikeLadder and the Codex team at NVIDIA HQ on how to roll it out to all of our employees. As a developer, our lives change every couple months. Now, the same is happening for all knowledge workers. I use Codex + 5.5 to complete features end to end, including updating linear tickets, reviewing PRs, and even tagging people in slack ;) But I also use it to summarize meetings from transcripts, read my email and give me action items, and create visualizations of data to help the company make informed decisions. OpenAI continues to uplevel everyone professionally with their tools, powered by NVIDIA hardware to accelerate them.

I’ve had access to Codex2 and GPT 5.5 for about 2 weeks now OpenAI just collison-installed all of NVIDIA lol They set up a lab like a Genius Bar so everyone could get set up w Codex 2 With the CLIs we’ve been building, non-technical coworkers seemed to have the biggest unlock Our stack: • We rolled out cloud VMs for every employee. Simple rule, agents get their own computers just like employees. If something goes wrong we can freeze it and get a stack trace. • Codex team has been super responsive. Codex 2 now supports any cloud vm, quickly picks up ssh config. Non-technical users just paste a prompt we gave them to edit ssh config • Internally built CLIs + those vetted by security get automatically loaded in cloud VMs. Teams wake up to new capabilities daily now. NVIDIA already moved fast now we’re rippin 🤙

Massive congrats to the team at @OpenAI 💚 Moving the frontier forward isn’t easy, and GPT-5.5 does exactly that, with impressive performance across agentic coding, research tasks, and beyond.

CUDA Live: Scaling HPC with Multi-GPU Communication Libraries x.com/i/broadcasts/1…

🌍💚 Happy Earth Day. We are spotlighting five projects where AI is helping solve some of the Earth’s greatest challenges. Check them out below . . . 🧵👇 ⛈️ Earth-2 climate simulation 🦧 Endangered species conservation ♻️ Robots sorting recyclables 🌊 Tsunami warnings 🌌 Earth observation insights

@gabeeegoooh Congrats on Images 2.0!

we needed those gpus

How to Contribute to Open Source - Live on NVIDIANemoClaw | Nemotron Labs x.com/i/broadcasts/1…

OpenShell v0.0.34 🧩 live sandbox policy updates 🖥️ install-vm now installs the gateway and VM driver, with new --driver-dir support 🛠️ sandbox get now shows the active runtime polic 🔒 supervisor seccomp improvements and HTTP normalization Update sandbox policy without restarting the runtime. github.com/NVIDIA/OpenShe…