Nemotron 3 Super.
A 120B open-weights text model (12B active) for agent workflows, tool use and retrieval, with up to 1M tokens of context and switchable reasoning.
2× Nvidia Q RTX 8000 · $0.51/hr.
Most-aggressive quantisation we have a working recommendation for. Lower precision = less VRAM = cheaper hardware, at a small accuracy cost.
Cheapest hosted endpoints.
Variants in the Nemotron family.
Nemotron 3.5 Lightning is NVIDIA's 30B mixture-of-experts text model, 3B acti...
A 550B open-weights text model (55B active) for complex agents, long-context ...
Nemotron 3.5 Content Safety is NVIDIA's 4B moderation model that labels promp...
A 120B open-weights vision-language model (12B active) that reads text, image...
NVIDIA Nemotron 3 Nano 30B A3B is a small language MoE model with highest com...
NVIDIA-Nemotron-Nano-9B-v2 is a large language model (LLM) trained from scrat...
Nemotron 3 Nano Omni is an open multimodal model built on a hybrid Mixture-of...
Cheapest hardware per quantisation.
Each row is one quantisation tier (the same weights compressed differently). Lower precision → lower VRAM → cheaper hardware, at the cost of small accuracy loss. $/hr is the cheapest rate on offer now. A multi-card setup uses a real instance of that size when one is rented; otherwise it is the single-card rate times the count.
| Quantisation | Cheapest GPU config | Total VRAM | Live $/hr | tokens/sec | |
|---|---|---|---|---|---|
|
FP16
FP16 — half precision (default)
|
384 GB |
$2.04/hr
est. 8 × single-card rate
Vast.ai · on-demand
|
— | Compare → | |
|
FP8
FP8 — 8-bit float (Hopper / Blackwell)
|
192 GB |
$1.02/hr
est. 4 × single-card rate
Vast.ai · on-demand
|
— | Compare → | |
|
INT4
INT4 — 4-bit integer (~4× VRAM saving)
|
96 GB |
$0.51/hr
est. 2 × single-card rate
Vast.ai · on-demand
|
— | Compare → |
What it costs per month across providers.
Estimate your monthly bill for Nemotron 3 Super across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.
Cheapest provider on the left.
Total monthly cost — input + output tokens combined.
Bill breakdown.
Rent the GPU instead of paying per token.
For an open-weights model like Nemotron 3 Super, you can rent a GPU and serve inference yourself. The math: cheapest GPU rental × 730 hours/month. Power is included in a rental's hourly price.
Assumes the GPU runs 24/7 at ~85% utilisation. If your traffic is bursty, you'll pay less for the API and probably more for the GPU (idle hours still cost rental). The breakeven analysis lives on the Self-host vs API breakeven tool.
About Nemotron 3 Super.
An open-weights text model from NVIDIA with 120B parameters, 12B of them active per token. It uses a LatentMoE design that combines Mamba-2, mixture-of-experts and attention layers, plus a built-in multi-token prediction head for faster decoding. Context runs up to 1M tokens, and reasoning can be switched on or off. NVIDIA aims it at agent workflows, long-context reasoning, tool use, retrieval and high-volume jobs such as IT ticket automation, and lists 8x H100 80GB as the minimum GPU setup.