Nemotron 3 Super 120B A12B

A 120B LatentMoE hybrid from NVIDIA activating 12B per token, combining Mamba-2, MoE and attention layers with Multi-Token Prediction, under the NVIDIA Nemotron Open Model License.

Params: 120B, 12B active
Context: 1M tokens
Cutoff: Jun 2025
Released: Mar 2026
License: NVIDIA Nemotron Open Model License

Visit Nvidia Compare vs other LLMs →

Hosted API price

Cheapest via Cloudflare · per 1M tokens

USD

Input: $0.50 /1M
Output: $1.50 /1M
Hosted by: 2 providers

Hosted API pricing

How we price

Order

Providers are listed cheapest input rate first, then by output rate, a tie going to the creator's own listing. A provider that lists the model without a published rate sits at the end.

Prices

Base-tier, on-demand rates in USD per 1M tokens, converted at ECB reference rates where a provider publishes in another currency. Cached-input and batch rates show only where a provider publishes them. "Cheapest" marks the one provider with the lowest input and output pair; a tie wears no badge.

Cost

Input rate times the input tokens plus output rate times the output tokens, at the monthly volume set above the table. Cached-input, batch, long-context and reasoning-token billing are not modelled.

Transparency and funding

Affiliates: Affiliate links are marked. We may earn a commission if you click them, but commissions never affect the order.

Every provider serving Nemotron 3 Super 120B A12B, cheapest input first. Set your monthly volume to see what each would bill.

Input M Output M

Every provider serving Nemotron 3 Super 120B A12B, with its price per 1M input and output tokens and what the volume above would cost
Provider	Input /1M	Output /1M	Cost at 10M in + 2M out	Link
Cloudflare	$0.50	$1.50	$8.00	Visit website
Nvidia Creator	Open weights, no hosted price			Visit website

Heads up: Base-tier, on-demand rates per 1M tokens; cached, batch and long-context tiers excluded. A provider may serve a shorter context or a quantized build than the creator's release. Verify before provisioning. More on how we price.

Estimated cost to self-host

Nemotron 3 Super 120B A12B needs about 68 GB of GPU memory at 4-bit with 32K context. The cheapest rental that fits is 4× RTX 3090 at about $662 a month.

That costs the same as roughly 994M tokens a month on Cloudflare's API. Self-hosting is more expensive below that volume.

Single node ≤ 8

What it costs to self-host Nemotron 3 Super 120B A12B at each precision: the memory it needs, the cheapest rental that holds it, that rental's month and the API volume it pays off at
Precision	Memory	Cheapest fit (32K context)(Full 1M context)	Cost /mo	Break-even vs API
4-bit INT4 / FP4	68 GB76 GB	4× RTX 3090	$662$662	994M tokens /mo994M tokens /mo
8-bit FP8 / INT8	122 GB130 GB	8× RTX 3090	$1,325$1,325	2B tokens /mo2B tokens /mo
16-bit FP16 / BF16	242 GB250 GB	8× RTX A6000	$3,398$3,398	5B tokens /mo5B tokens /mo

Estimates based on median on-demand rates for Nvidia GPUs. Memory is weights plus KV cache for one request, using FP8 KV cache where supported and FP16/BF16 otherwise. Break-even assumes a 5:1 input-to-output ratio. No guarantee of runtime support, usable performance, or that a matching quantized build exists. Pricing methodology.

Similarly priced models

The models nearest Nemotron 3 Super 120B A12B by blended rate, each at its own cheapest provider.

Models priced nearest Nemotron 3 Super 120B A12B per 1M tokens, each at its cheapest provider
Model	Blended /1M	Input /1M	Output /1M	Context	Cutoff	vs Nemotron 3 Super 120B A12B
GPT-4.1 Mini OpenAI	$0.60	$0.40	$1.60	1M	Jun 2024	−10%
Qwen2.5-72B Instruct Alibaba Cloud		$0.62		33K
Gemini 2.5 Flash Google Cloud	$0.6667	$0.30	$2.50	1M	Jan 2025	0%
Gemini 3.5 Flash-Lite Google Cloud	$0.6667	$0.30	$2.50	1M	Mar 2026	0%
Mistral Large 3 Mistral	$0.6667	$0.50	$1.50	256K		0%
Nemotron 3 Super 120B A12B This model Nvidia	$0.6667	$0.50	$1.50	1M	Jun 2025
Qwen3.5-Plus Alibaba Cloud	$0.6683	$0.36	$2.21	1M		0%
DeepSeek V3.2 DeepSeek	$0.6833	$0.50	$1.60	131K		+3%
Qwen3.6-27B Alibaba Cloud	$0.7564	$0.4126	$2.4754	262K		+13%
DeepSeek V3 DeepSeek		$0.85		128K
GLM-4.7 Z.AI	$0.8667	$0.60	$2.20	205K		+30%

USD per 1M tokens at each model's cheapest listed provider. Blended is the cost of 10M input plus 2M output tokens, spread over the 12M.

Runpod

Spin Up a GPU in Seconds - No Setup Required

Deploy AI Models Globally in 30+ Regions

Scale from 0 to 1000 GPUs on Demand

PyTorch, Tensorflow & Docker Ready

Learn more

Capabilities

What Nemotron 3 Super 120B A12B accepts and can do, as published by Nvidia.

Text Accepts and generates natural-language text.
Reasoning Thinks before it answers, always on or as a switchable mode.
Function calling Connect to external tools, APIs, and systems.
Structured output Return responses in structured formats like JSON.

Common questions

What is Nemotron 3 Super 120B A12B good for?

Agent and high-volume work such as IT ticket automation. Open weights under the NVIDIA Nemotron Open Model License, servable on a single high-memory GPU.

When is Nemotron 3 Super 120B A12B not a good fit?

Image input: it is text only. Memory follows the full 120B parameters, not the 12B active, and no maximum output is published.

What is the cheapest way to run Nemotron 3 Super 120B A12B?

Hosted, unless you push serious volume. Cloudflare charges $0.50 in / $1.50 out per 1M tokens. The cheapest rental that fits is 4x RTX 3090 at $662 a month, which costs the same as about 994M tokens a month on that API.

How does Nemotron 3 Super 120B A12B compare with Nemotron 3.5 Super VL 120B A12B?

NVIDIA says Nemotron 3.5 Super VL starts from the Nemotron 3 Super base, adds a vision encoder for images, documents and video, and was further trained by distillation from Nemotron 3 Ultra and on software-engineering data. Same 1M-token context. Nemotron 3 Super 120B A12B costs 233% more per 1M input tokens and 200% more per 1M output tokens than Nemotron 3.5 Super VL 120B A12B, at each model's cheapest listed provider. It came out 7 months earlier, takes no image and video input, which Nemotron 3.5 Super VL 120B A12B does and is licensed under NVIDIA Nemotron Open Model License against OpenMDW 1.1.

Can I self-host Nemotron 3 Super 120B A12B?

Yes. The weights are released under the NVIDIA Nemotron Open Model License. At 4-bit it needs about 68 GB of GPU memory, which starts at roughly $662 a month on the cheapest rental that fits.