frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Nvidia Nemotron 3 Family of Models

https://research.nvidia.com/labs/nemotron/Nemotron-3/
101•ewt-nv•1d ago

Comments

Y_Y•1d ago
Wow, Nvidia keepson pushing the frontier of misleading benchmarks
pants2•1d ago
If it's intelligence + speed you want, nothing comes close to GPT-OSS-120B on Cerebras or Groq.

However, this looks like it has great potential for cost-effectiveness. As of today it's free to use over API on OpenRouter, so a bit unclear what it'll cost when it's not free, but free is free!

https://openrouter.ai/nvidia/nemotron-3-nano-30b-a3b:free

viraptor•1d ago
> nothing comes close to GPT-OSS-120B on Cerebras

That's temporary. Cerebras speeds up everything, so if Nemotron is good quality, it's just a matter of time until they add it.

credit_guy•22h ago
That's unlikely. Cerebras doesn't speed up everything. Can it speed up everything? I don't know, I'm not an insider. But does it speed up everything? That is evidently not the case. Their page [1] lists only 4 production models and 2 preview models.

[1] https://inference-docs.cerebras.ai/models/overview

red2awn•1d ago
Very interesting release:

* Hybrid MoE: 2-3x faster than pure MoE transformers

* 1M context length

* Trained on NVFP4

* Open Source! Pretraining, mid-training, SFT and RL dataset released (SFT HF link is 404...)

* Open model training recipe (coming soon)

Really appreciate Nvidia being the most open lab but they really should make sure all the links/data are available on day 0.

Also interesting that the model is trained in NVFP4 but the inference weights are FP8.

bcatanzaro•7h ago
The Nano model isn’t pretrained in FP4, only Super and Ultra are. And posttraining is not in FP4, so the posttrained weights of these models are not native FP4.
wcallahan•22h ago
I don’t do ‘evals’, but I do process billions of tokens every month, and I’ve found these small Nvidia models to be the best by far for their size currently.

As someone else mentioned, the GPT-OSS models are also quite good (though I haven’t found how to make them great yet, though I think they might age well like the Llama 3 models did and get better with time!).

But for a defined task, I’ve found task compliance, understanding, and tool call success rates to be some of the highest on these Nvidia models.

For example, I have a continuous job that evaluates if the data for a startup company on aVenture.vc could have overlapping/conflated two similar but unrelated companies for news articles, research details, investment rounds, etc… which is a token hungry ETL task! And I recently retested this workflow on the top 15 or so models today with <125b parameters, and the Nvidia models were among the best performing for this type of work, particularly around non-hallucination if given adequate grounding.

Also, re: cost - I run local inference on several machines that run continuously, in addition to routing through OpenRouter and the frontier providers, and was pleasantly surprised to find that if I’m a paying customer of OpenRouter otherwise, the free variant there from Nvidia is quite generous for limits, too.

btown•23m ago
Would you mind sharing what hardware/card(s) you're using? And is https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B... one of the ones you've tested?
max002•13h ago
Im upvoting, im happy to finally see open source model with commercial use from Nvidia as most of the models ive been checking from you guys couldnt be used in commercial settings. Bravo Nvidia!
kristianp•1h ago
The article seem to focus on the nano model. Where are the details of the larger ones?
shikon7•1h ago
> We are releasing the Nemotron 3 Nano model and technical report. Super and Ultra releases will follow in the coming months.
jtbayly•53m ago
Any chance of running this nano model on my Mac?

Prediction: AI will make formal verification go mainstream

https://martin.kleppmann.com/2025/12/08/ai-formal-verification.html
69•evankhoury•1h ago•30 comments

alpr.watch

https://alpr.watch/
537•theamk•5h ago•268 comments

No Graphics API

https://www.sebastianaaltonen.com/blog/no-graphics-api
262•ryandrake•3h ago•40 comments

GPT Image 1.5

https://openai.com/index/new-chatgpt-images-is-here/
218•charlierguo•4h ago•117 comments

Ty: A fast Python type checker and LSP

https://astral.sh/blog/ty
78•gavide•1h ago•11 comments

MIT professor shot at his Massachusetts home dies

https://www.bbc.com/news/articles/cly08y25688o
26•mosura•34m ago•0 comments

40 percent of fMRI signals do not correspond to actual brain activity

https://www.tum.de/en/news-and-events/all-news/press-releases/details/40-percent-of-mri-signals-d...
357•geox•8h ago•157 comments

Mozilla appoints new CEO Anthony Enzor-Demeo

https://blog.mozilla.org/en/mozilla/leadership/mozillas-next-chapter-anthony-enzor-demeo-new-ceo/
359•recvonline•8h ago•526 comments

Thin desires are eating life

https://www.joanwestenberg.com/thin-desires-are-eating-your-life/
228•mitchbob•21h ago•86 comments

The World Happiness Report is beset with methodological problems

https://yaschamounk.substack.com/p/the-world-happiness-report-is-a-sham
66•thatoneengineer•22h ago•84 comments

Writing a blatant Telegram clone using Qt, QML and Rust. And C++

https://kemble.net/blog/provoke/
56•tempodox•6h ago•30 comments

Sega Channel: VGHF Recovers over 100 Sega Channel ROMs (and More)

https://gamehistory.org/segachannel/
195•wicket•9h ago•27 comments

GitHub will begin charging for self-hosted action runners on March 2026

https://github.blog/changelog/2025-12-16-coming-soon-simpler-pricing-and-a-better-experience-for-...
368•nklow•4h ago•147 comments

Chat-tails: Throwback terminal chat, built on Tailscale

https://tailscale.com/blog/chat-tails-terminal-chat
11•nulbyte•1h ago•1 comments

Nvidia Nemotron 3 Family of Models

https://research.nvidia.com/labs/nemotron/Nemotron-3/
102•ewt-nv•1d ago•12 comments

Show HN: Sqlit – A lazygit-style TUI for SQL databases

https://github.com/Maxteabag/sqlit
86•MaxTeabag•1d ago•9 comments

Artie (YC S23) Is Hiring Senior Enterprise AES

https://www.ycombinator.com/companies/artie/jobs/HyaHWUs-senior-enterprise-ae
1•j-cheong•5h ago

Context: Odin’s Most Misunderstood Feature

https://www.gingerbill.org/article/2025/12/15/odins-most-misunderstood-feature-context/
25•davikr•1d ago•0 comments

Letta Code

https://www.letta.com/blog/letta-code
16•ascorbic•1h ago•1 comments

Creating custom yellow handshake emojis with zero-width joiners

https://blog.alexbeals.com/posts/custom-yellow-handshake-emojis-with-zero-width-joiners
44•dado3212•22h ago•2 comments

Show HN: Deterministic PCIe Diagnostics for GPUs on Linux

https://github.com/parallelArchitect/gpu-pcie-diagnostic
6•gpu_systems•1h ago•1 comments

Rust GCC back end: Why and how

https://blog.guillaume-gomez.fr/articles/2025-12-15+Rust+GCC+backend%3A+Why+and+how
150•ahlCVA•8h ago•70 comments

How geometry is fundamental for chess

https://lichess.org/@/RuyLopez1000/blog/how-geometry-is-fundamental-for-chess/h31wwhUX
43•fzliu•5d ago•16 comments

30 Years of <Br> Tags

https://www.artmann.co/articles/30-years-of-br-tags
124•FragrantRiver•3d ago•25 comments

Pizlix: Memory Safe Linux from Scratch

https://fil-c.org/pizlix
55•nullbyte808•2d ago•17 comments

Vibe coding creates fatigue?

https://www.tabulamag.com/p/too-fast-to-think-the-hidden-fatigue
118•rom16384•3h ago•118 comments

Purrtran – ᓚᘏᗢ – A Programming Language for Cat People

https://github.com/cmontella/purrtran
213•simonpure•3d ago•31 comments

Full Unicode Search at 50× ICU Speed with AVX‑512

https://ashvardanian.com/posts/search-utf8/
178•ashvardanian•1d ago•69 comments

Confuse some SSH bots and make botters block you

https://mirror.newsdump.org/confuse-some-ssh-bots.html
38•Bender•5d ago•14 comments

The Beauty of Dissonance

https://www.plough.com/en/topics/culture/music/the-beauty-of-dissonance
7•tintinnabula•3d ago•0 comments