gordana neskovic

After much reflection, I have decided to resign from my position as Director of the National Counterterrorism Center, effective today. I cannot in good conscience support the ongoing war in Iran. Iran posed no imminent threat to our nation, and it is clear that we started this war due to pressure from Israel and its powerful American lobby. It has been an honor serving under @POTUS and @DNIGabbard and leading the professionals at NCTC. May God bless America.

💪

NVIDIA Announces NemoClaw for the OpenClaw Community nvidianews.nvidia.com/news/nvidia-an…

Tokenization can be a #TTFT bottleneck for long-context #LLMs. That's why we worked with the @NVIDIA Dynamo team to built fastokens: ⚡ Up to 17× faster on 50K+ tokens 📉 Reduced TTFT by up to 40% Read more: crusoe.ai/resources/blog… #NVIDIAGTC #GTC2026 #AIInfrastructure

We're excited to be day-0 launch partners for NVIDIA Nemotron 3 Super! You can try it now on Baseten, or read @rapprach's blog to learn more about the new model: baseten.co/blog/introduci…

Try it now for free: openrouter.ai/nvidia/nemotro…

Introducing NVIDIA Nemotron 3 Super 🎉 Open 120B-parameter (12B active) hybrid Mamba-Transformer MoE model Native 1M-token context Built for compute-efficient, high-accuracy multi-agent applications Plus, fully open weights, datasets and recipes for easy customization and deployment. 🧵

🚀 Day 0 support for Nvidia's Nemotron 3 Super! We're excited to support open source models that push the frontier of model intelligence, cost, and latency Try it out in deepagents today!

NVIDIA Nemotron 3 Super is built for multi-agent systems. • 120B MoE model • 1M context • Multi-token prediction But agent-grade models need agent-grade infrastructure 👇

🎉 Congrats on the launch of Nemotron 3 Super from @nvidia! Day-0 support is now live in SGLang. A 120B hybrid MoE (only 12B active params) built for high-efficiency multi-agent systems: ⚡️ 5x higher throughput vs. the previous Nemotron Super 🧠 Multi-Token Prediction (MTP): predicts multiple tokens per forward pass 🎯 Thinking Budget: dial in optimal accuracy with minimum reasoning tokens 📜 1M-token context: keep full conversation history and plan state in view across long workflows 🏆 2x higher accuracy on Artificial Analysis Intelligence Index vs. the previous Nemotron Super Cookbook: cookbook.sglang.io/autoregressive… Blog: blogs.nvidia.com/blog/nemotron-… Run it now with SGLang!

🎉 Congrats to @nvidia on the release of Nemotron 3 Super — day-0 support in vLLM v0.17.1! Verified on NVIDIA GPUs. 120B hybrid MoE, only 12B active at inference. Big upgrades over the previous Nemotron Super: - 5x higher throughput - 2x higher accuracy on Artificial Analysis Intelligence Index - Multi-Token Prediction (MTP) for faster long-form generation - Configurable thinking budget — dial accuracy vs token cost per task - 1M token context window Supports BF16, FP8, and NVFP4. Fully open: weights, datasets, recipes. Blog: vllm.ai/blog/nemotron-… 🤝 Thanks @NVIDIAAIDev Nemotron team and vLLM community contributors!

Day Zero fine-tuning & hosting support for Nemotron 3 Super by @nvidia is now live Fine-tune on real production traces & deploy on high-performance infrastructure optimized for Nemotron 3 Super Your data, your weights, your performance edge Learn more: inference.net/blog/nemotron-…

The system doesn't just manufacture consent. It also manufactures dissent and decides which dissidents you're allowed to hear. Michael Parenti published "Inventing Reality" in 1986. Chomsky's "Manufacturing Consent" came out in 1988. Chomsky cited Parenti exactly once, for an article, not the book that preceded his own on the same subject. Same thesis: media serves power. But the books are not the same. On "conspiracy": Chomsky dismisses structural analysis that names actors as "conspiracy theorism." Parenti: "The alternative to a conspiracy theory is an innocence theory... the CIA is by definition a conspiracy." Chomsky wanted you to protest... abstractly. Never name the names. Right-wing anticommunism is obvious. But they needed someone to attack socialism from the left, to gut-punch communism from inside the tent. Enter Chomsky. Chomsky equates Stalin with Goebbels, called Leninism "counterrevolution." Parenti documented the real gains: full employment, free healthcare, free education, free housing. He wrote about how the existence of the USSR forced concessions in the capitalist countries: social programs, worker rights, formal decolonization. Capital had to compete with an alternative Chomsky made sure you wouldn't defend. On empire: Chomsky framed US foreign policy as blunders and mistakes. Parenti said "the Iraq war has not been a mistake", it succeeded for the class it serves. Your taxes, your children's lives. Their profits, their reconstruction contracts. Externalized costs, privatized gains. On intervention: Chomsky supported the NATO no-fly zone over Libya. Parenti opposed it, listed what Libyans would never see again after "liberation." On elections: Chomsky did the "lesser evil" dance every four years, vote Democrat in swing states. Parenti broke with Bernie Sanders over the bombing of Yugoslavia. Chomsky got MIT, The Guardian, documentaries, Jacobin. Parenti barely got lecture halls and public access TV. Parenti was the one they didn't want you listening to, so they put Chomsky in front of you.

Thanks @NVIDIAAI for inviting us to Dynamo Day! We're active users of Dynamo, iterating on it in production for performance gains like 50% lower TTFT and 34% lower TPOT, and regularly shipping our work back to the community. Read some of our highlights from Dynamo Day and working with NVIDIA Dynamo here: baseten.co/blog/nvidia-dy…

Quantization-aware distillation (QAD) delivers near-BF16 accuracy with NVFP4! Amazing work led by @huizi_mao

Big News 🎉 NVIDIA’s AI Planner implementation has been selected as a finalist for the 2026 Franz Edelman Award, also known as “The Nobel Prize” of Operations Research. Our innovative analytics solution helped achieve incredible results: ➡️ 83% reduction in delayed orders ➡️ Scenario analysis compressed from 5 days to just 1 day ➡️ 6X expansion in planning capacity 📍 Join us April 13 at the INFORMS Analytics Conference Gala finals as we celebrate the innovators driving real-world impact through analytics and operations research ➡️ nvda.ws/4a3bU5h

PersonaPlex-7B just hit #1 on @huggingface 🎉 Part of the NVIDIA Nemotron collection of models, PersonaPlex can listen and speak at the same time for fast, natural-feeling conversations. Plus, it’s open so you can tune personas, clone voices, and run it locally. Check it out 👉 nvda.ws/3MbHWUJ Congratulations to our speech research team 🙌

🎉 We are thrilled to announce that our paper on CoDeC (Contamination Detection via Context) has been accepted to #ICLR2026!