Nebius

Building AI Agents is only half the challenge. The other half is knowing whether it works and getting it into production. Introducing the Agentic AI course from Nebius and NVIDIA, a free, hands-on program built around one project: a deep research agent you take from the first API call to deployment. Along the way, you'll build on NVIDIA Nemotron Open Models via Nebius Token Factory, give your agent live web access with Tavily, and learn how to evaluate what you've built before it ships. Alex Hanley, Solutions Architect for Nebius Token Factory, walks through what's inside. Enroll for free: https://lnkd.in/gicg_6ME

We're heading to GTC Berlin as a Diamond Sponsor. 💎 From October 20 to 22, join us to explore what it takes to run AI in production, from the cloud infrastructure behind demanding workloads to what we’re building with NVIDIA. You'll also be able to catch us on stage! Danila Shtan, Chief Technology Officer, Andrew Eliseev, Head of Retail & Commerce AI, and Nir Chervoni, Head of Security Products are all presenting. Find us at Booth 3013 for live demos, swag, and conversations with our team. Beyond the show floor, you can also join us at Nebius Haus, our dedicated space just off the show floor. Get the details: https://lnkd.in/eCHtkdnR

️Over half of our cities down: Tokyo, Da Nang, Seoul, Kuala Lumpur, Singapore, Tel Aviv, London, Copenhagen, Stockholm, Taipei, NYC, Warsaw, Mexico City, and Amsterdam. At every stop, developers, founders, and AI builders have come together to exchange ideas, get hands-on technical support, and make real progress on what they are building. A 20-city global tour takes a serious collective effort. Thank you to Tavily, NVIDIA, and every local host, team member, speaker, and builder who has made it possible. Check out the highlights from one of our most recent spots: London! Still ahead: Berlin, Toronto, Paris, Boston, San Francisco, and Los Angeles. Let’s build. Find your nearest stop: https://lnkd.in/ebcs5Fbj

What does it take to build AI that understands clinical conversations? Join Nebius on October 1 for Building and Evaluating Clinical Speech AI, a practical webinar supported by NVIDIA and J.P. Morgan. Through a live demo and Q&A with engineers, we’ll explore how to: - Deploy NVIDIA Nemotron Speech on Nebius. - Adapt speech recognition to medical terminology. - Evaluate accuracy and safety for clinical use. - Plan for production with data privacy in mind. For developers, ML engineers, product teams and technical leaders building AI-powered healthcare applications. No prior experience with Nemotron is required. Register now: https://lnkd.in/gRYzwamc

Coming to GTC Berlin? Save 20% on your conference pass with our code ref-spo-265118. We'll be on the show floor at Booth # 3013 as a Diamond Sponsor, plus running community events all week through Nebius Build Week. Grab your pass: https://lnkd.in/emWVN6GJ

What happens when long-running agentic workloads outgrow GPU memory? WEKA and Nebius ran an eight-hour benchmark on identical NVIDIA HGX B300 hardware, changing only the KV cache configuration. With WEKA’s Augmented Memory Grid pooling local NVMe across the cluster, the system served 2.4× more requests, recomputed 3.2× fewer prefill tokens and maintained median time to first token near one second throughout the test. See the full results: https://lnkd.in/eP-RvHB5

Issue 2 of Build with Nebius is live. It's built for anyone assembling an AI stack right now. Headlining it: the new AI Builder Program. $400+ in credits and discounts, runnable cookbooks, free courses, and office hours with engineers. It's free to join and the benefits are endless. Also inside: a free agentic AI course built with NVIDIA and Tavily that walks you through building a deep research agent from scratch, three real builds from the Nebius Academy community (including a paper reviewer that backs 92% of its grades with a verified quote in 38 seconds), and a Tavily breakdown of what self-hosted web search actually costs you. Subscribe so issue 3 lands straight in your inbox: https://lnkd.in/eX6jzgT5

"Nebius is unquestionably an industry leader with strong offerings in every category." That's SemiAnalysis' independent verdict on Nebius AI Cloud. SemiAnalysis just gave Nebius a Platinum rating, its highest tier for AI cloud providers. The rating followed live testing on production clusters over several months: burn-ins, storage under real load, injected hardware failures, full recovery drills. Real racks, real conditions. When you're choosing where to run production AI, this is what's actually at stake: less time firefighting infrastructure, more time on the workloads you actually care about. That's what independent testing like this validates. See what SemiAnalysis found: https://lnkd.in/efZZetX2

Today we're introducing spot pricing for preemptible VMs on Nebius. Until now, the discount on preemptible capacity has been a number we set. From October 8, the price will be calculated dynamically from available capacity and demand, for each GPU type in each region, and it will always stay below the on-demand rate. You choose how to work with it. Set a maximum price per GPU-hour and run when the price is at or below it, or follow the price and prioritize keeping your workload running. Both settings are available today, and price history will be published so you can see how the price moves before you decide. If you already run preemptible VMs, choose one of the two before October 8. Read how it works: https://lnkd.in/eKEm4aak

Nebius is coming to Zurich! 🇨🇭 We’re growing our presence in one of Europe's most vibrant tech hubs and building our Agentic Search team locally, anchored around the Tavily product. The team will work alongside Zurich’s research community, with openings across AI research, engineering, and product. On October 7, we’re hosting a technical meetup all about Agentic Search. Come swap ideas, meet other AI professionals, and join the conversation: https://luma.com/mgmq5ipe  #Zurich #HiringAgenticSearch #AIInfrastructure