Nemotron 3 Ultra is coming this week. ⌛️

Proud to be part of this coalition!

free access?

let's goooo!!! very excited and honored to be part of the coalition + contribute to the truly open frontier :)

why not compare with qwen 3.6 and Kimi k 2.6? 3.5 and 2.5 are old, update your charts

$NVDA investors are eating good

550B open model with 5x faster inference and 30% lower cost? NVIDIA is really going all-in on open frontier AI. Can't wait to test it!

Could NVIDIA’s Nemotron 3 Ultra finally birth the first generation of truly persistent AI companions, minds that remember every conversation, plan across weeks, and evolve with us over a million-token horizon?

I really hope you guys release Ultra in lockstep with a solid dgx spark implementation and optimization plan this time. I spent months optimizing Nemotron 3 super with my spark because you released very little guidance for the spark when it dropped

Atbash Inside

these guys get as much compute as they want and can't even make better than most of ai labs

GOATs🫡

Big money rotating out of oil and into AI stocks. Classic sector shift in action.

It's probably time to implement Nemotron into our system.

Nemotron 4 will be absolute 🔥

Nvidia capping their big drop at 128G unified memory right as Apple is about to drop the M5 Ultra Studio 512GB is utterly retarded.

Nop, they are lying or maybe recording was old... They do not compare agains world's best openmodels. I do not see deepseek there, i do not see qwen3.7 there but qwen3.5 (what a joke)

重点看推理延迟和实际吞吐。企业落地更关心成本与稳定性，期待详细测试数据。

Hope it free and speed

Hyped.

This in context with Vera Rubin

Making people extinct, run buy

550B gets attention. I'm more interested in whether it's fast and cheap enough to leave running. That's the line for agents.

Great work!

We not even into first week of the month yet and NVIDIA is already ahead of most companies out there again

N1X cool

Nice.

@grok Which "best open models" is the 5×/30% comparison against, and on what benchmark?

We are not a car! We are not a car!

🔥 we are ready for it!

老黄的皮夹克依旧的亮

I am super excited for this!!

550B参数55B激活，5x更快30%更便宜，老黄这是来定标准不是来参选。芯片厂搞开源模型最大的底气就是我有算力我有卡，模型白送你也得在我卡上跑。商业闭环玩明白了。

JUST IN: Jensen Huang announced the launch of Nemotron 3 Ultra, NVIDIA’s latest open AI model, during activities surrounding COMPUTEX 2026. The model is part of NVIDIA’s broader Nemotron 3 family focused on agentic AI, reasoning systems, and large-scale autonomous workflows