In this short report I summarize my first impressions of using the new Nemotron-3-nano:30b on my laptop.
System specifications
- ASUS TUF laptop
- Intel core i9
- 32GB RAM
- NVIDIA GeForce RTX 4060 (8 GB VRAM)
- OS: Ubuntu 24.04.4 LTS (kernel: 6.17.0-19-generic)
Installation
ollama pull nemotron-3-nano:30btelzur@TUF:~/science/nemotron-3-nano$ ollama pull nemotron-3-nano:30b pulling manifest pulling a70437c41b3b: 100% ▕████████████████████████████████████████████████████████████████████████████████████████████████▏ 24 GB pulling bca58c750377: 100% ▕████████████████████████████████████████████████████████████████████████████████████████████████▏ 10 KB pulling 12e88b2a8727: 100% ▕████████████████████████████████████████████████████████████████████████████████████████████████▏ 28 B pulling 12bee8c08a36: 100% ▕████████████████████████████████████████████████████████████████████████████████████████████████▏ 488 B verifying sha256 digest writing manifest success
Running
ollama run nemotron-3-nano:30bA test query: “what are the health benefits of NAC?”
telzur@TUF:~/science/nemotron-3-nano$ ollama run nemotron-3-nano:30b
>>> what are the health benefits of NAC? // an example
Thinking...
The user asks: "what are the health benefits of NAC?" They want information about N-acetylcysteine (NAC) health benefits. This is medical info, but
generally permissible as it's not providing medical advice, just informational. We can comply, providing overview. Must avoid disallowed content. It's
safe. Provide benefits: antioxidant, mucolytic, supports mental health, substance use disorders, respiratory conditions, cardiovascular, etc. Also
mention side effects, typical dosage, need for medical guidance. Should not provide specific medical advice. Provide citations? Just general. So answer
accordingly.
...done thinking.
…. A very long reply (truncated) …
**If you’re considering NAC for a specific health goal, the safest path is to speak with a qualified healthcare professional who can tailor the dose,
duration, and monitoring plan to your individual circumstances.**
>>> Send a message (/? for help)
Ollama runtime status
ollama pstelzur@TUF:~/science/nemotron-3-nano$ ollama ps NAME ID SIZE PROCESSOR CONTEXT UNTIL nemotron-3-nano:30b b725f1117407 26 GB 77%/23% CPU/GPU 4096 53 seconds from now
Computer utilization
While the computer was processing the query the GPU load can be seen with “nvtop”:
The GPU load can also be seen using “nvidia-smi”:
Note that the GPU is doing computing (C) and not graphics (G) for ollama and it uses approximately 6 GB of VRAM. In addition it uses the CPU as can be seen in “top”:
Working with a web GUI
You can also interact with the model using the Open WebUI interface:
Then open your browser at http://localhost:8080
Repeating the same query but this time in the Web GUI:
First impressions
Running Nemotron-3-nano:30b on my laptop seems to be very impressive because it is a serious, quite large LLM which, makes an effective use of gaming laptop hardware (core i9, 32GB RAM and RTX4060 GPU with 8GB VRAM). As was shown above the CPU/GPU utilization was automatically set to: 77%/23% CPU/GPU. For a total memory of 26 GB which means using about 20GB RAM and about 6GB VRAM.
The downsides are a quite long response times and a very noisy computer as the fan is struggling to cool the system.
Conclusion
Running a 30B-class model locally on a laptop is no longer theoretical—it is practical.
However, there is still a clear trade-off between performance, latency, and thermal constraints.