Nvidia Nemotron

“they need to use nemotron for true sovereignty and stop paying the tax”

“It's OK, but makes an error every few sentences”

“Nemotron is TERRIBLE, and purposefully so.”

“There are a lot of open-research on pre-training, post-training and RL data mixtures and sourcing.”

“Not just open weights but open training data too and full tutorials on how to use them”

“Nvidia keeps contributing to the open ecosystem with nemotron”

“Nvidia is that most promotes open models and contributes significantly through Nemotron and various hardware-level optimizers.”

“not especially great, compared to Chinese ones”

“Nvidia releases some of the most open open weights models, Nemotron 3, which have the full training code open”

“they've consistently demonstrated the most openness when it comes to model training, datasets, and research”

“They also have very good code and playbooks for actually doing a fine-tune, CPT, etc.”

“Nemotron is a very nice model with an excellent license as well.”

“Nemotron 3 also introduced LatentMoE, which was adopted by Kimi K3”

“Each of the Nemotron 3 models were briefly competitive when they launched, but never exceptional”

“I've had some pretty decent results with Hy3 and Nemotron too, but they're not the same”

“I think the spiritual successor is the Nemotron 3 series, although they also are getting a bit long in the tooth”

“NVIDIA's Nemotrons are decent bases for finetuning or continued pretraining, esp thanks to good documentation and tooling”

“nemotron's time to first token is INSANELY fast. It's become a legit option for voice use cases”

“I have never used it enough to exhaust it but on openrouter I burnt a good 100M tokens through it”

“Nemotron is not worth running on any sillicon”

“it is pretty good at instruction following and has extremely fast decode”

“it performed extremely poorly, essentially unusable”

“better models that are state of the art now like Nemotron and Parakeet”

“They're releases so far have been kind of lackluster compared to Qwen and other Chinese models.”

“Its entire training pipeline is open-sourced, while other vendors typically only release the model weights.”

“Yesterday I used the latest one on OpenRouter and it was bad - worse than StepFun”

“Nemotron is the strongest model (on most benchmarks) that has its full training pipeline and most of the data open.”

“Nemotron only releases portions of some of their datasets”

“Nemotron is genuinely open source at least at the smaller sizes.”

“lots of folks dramatically underestimate their frontier models (Nemotron family), and they license those models (free)”

“Nemotron 3 is amazing. 60 tokens/s on my 128GB Nvidia GB10”

“They hav become a leading open source training lab themselves and they’re releasing the best training data, training tooling, and models at this point.”

“in my testing, they are the best performing models on low-end consumer hardware in both terms of speed and accuracy”

“very useful for coding if I prepare a one-shot prompt with directions and relevant code”

“I had mixed results with the new Nemotron”

“it's good enough for my tasks that I have no use for cloud models”

“60t/s on the 8 bit MLX on an M4 Pro with 64GB of RAM.”

“It completely mopped the floor with other models for my task (text style transfer)”

“I’ve found these small Nvidia models to be the best by far for their size currently.”

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