Llama Nemotron Super 49B TPS calculator

Open weights NVIDIA 49B parameters March 2025

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

93 of 818 cards that can run it

Smallest card that fits

Radeon PRO V710

28 GB · Q3_K_M · 9.2 tok/s

Fastest card

B200

69.2 tok/s · 180 GB

Which GPUs can run Llama Nemotron Super 49B?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

93 cards match

Calculating

					Needs	Quantisation	Fit
69.2 tok/s 41–111 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	53.2 GB	Q8_0	Comfortable
69.2 tok/s 41–111 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	53.2 GB	Q8_0	Comfortable
55.2 tok/s 33–88 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	53.2 GB	Q8_0	Comfortable
55.2 tok/s 33–88 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	53.2 GB	Q8_0	Comfortable
44.2 tok/s 27–71 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	53.2 GB	Q8_0	Comfortable
42.3 tok/s 25–68 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	53.2 GB	Q8_0	Comfortable
42.3 tok/s 25–68 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	53.2 GB	Q8_0	Comfortable
40.5 tok/s 24–65 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	53.2 GB	Q8_0	Comfortable
39.7 tok/s 24–64 · low confidence	DRIVE A100 PROD NVIDIA	32 GB	1,870 GB/s	May 2020	27.5 GB	IQ4_XS	Tight
39.7 tok/s 24–64 · low confidence	GRID A100A NVIDIA	32 GB	1,870 GB/s	May 2020	27.5 GB	IQ4_XS	Tight
38.0 tok/s 23–61 · low confidence	GeForce RTX 5090 NVIDIA	32 GB	1,790 GB/s	Jan 2025	27.5 GB	IQ4_XS	Tight
38.0 tok/s 23–61 · low confidence	GeForce RTX 5090 D NVIDIA	32 GB	1,790 GB/s	Jan 2025	27.5 GB	IQ4_XS	Tight
35.9 tok/s 22–57 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	53.2 GB	Q8_0	Comfortable
35.9 tok/s 22–57 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	53.2 GB	Q8_0	Comfortable
35.9 tok/s 22–57 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	53.2 GB	Q8_0	Comfortable
34.1 tok/s 20–54 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	53.2 GB	Q8_0	Comfortable
31.1 tok/s 19–50 · low confidence	A100 PCIe 40 GB NVIDIA	40 GB	1,560 GB/s	Jun 2020	30.3 GB	Q4_K_M	Tight
31.1 tok/s 19–50 · low confidence	A100 SXM4 40 GB NVIDIA	40 GB	1,560 GB/s	May 2020	30.3 GB	Q4_K_M	Tight
31.1 tok/s 19–50 · low confidence	A800 PCIe 40 GB NVIDIA	40 GB	1,560 GB/s	Nov 2022	30.3 GB	Q4_K_M	Tight
29.0 tok/s 17–46 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	53.2 GB	Q8_0	Comfortable
29.0 tok/s 17–46 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	53.2 GB	Q8_0	Comfortable
29.0 tok/s 17–46 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	53.2 GB	Q8_0	Comfortable
29.0 tok/s 17–46 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	53.2 GB	Q8_0	Comfortable
29.0 tok/s 17–46 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	53.2 GB	Q8_0	Comfortable
24.0 tok/s 14–38 · low confidence	Tesla PG500-216 NVIDIA	32 GB	1,130 GB/s	Nov 2019	27.5 GB	IQ4_XS	Tight

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: NVIDIA
Organisation type: Industry
Country: United States of America
Published: 18 March 2025
Authors: Akhiad Bercovich, Itay Levy, Izik Golan, Mohammad Dabbah, Ran El-Yaniv, Omri Puny, Ido Galil, Zach Moshe, Tomer Ronen, Najeeb Nabwani, Ido Shahaf, Oren Tropp, Ehud Karpas, Ran Zilberstein, Jiaqi Zeng, Soumye Singhal, Alexander Bukharin, Yian Zhang, Tugrul Konuk, Gerald Shen, Ameya Sunil Mahabaleshwarkar, Bilal Kartal, Yoshi Suhara, Olivier Delalleau, Zijia Chen, Zhilin Wang, David Mosallanezhad, A…

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Language
Task: Language modeling/generation, Question answering, Quantitative reasoning, Code generation, Neural Architecture Search - NAS
Base model: Llama 3.3 70B

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 49B
Training data: 100,000,000,000 tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

How it was established: Operation counting
Fine-tuning compute: 2.9 × 10²² FLOP

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (restricted use)
Training code: Unreleased
Hugging Face: nvidia

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Likely

Sources

Where this record came from and when it was last checked.

Reference: Super is 49B distilled from Llama 3.3 70B for best accuracy with highest throughput on a data center GPU.
Last updated: 28 November 2025

The extremes

The ten fastest GPUs for Llama Nemotron Super 49B

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run Llama Nemotron Super 49B

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

What you need to run it

Minimum card

Radeon PRO V710

Memory needed

24.6 GB

Fastest

69.2 tok/s

With 49B parameters, Llama Nemotron Super 49B lands in the range a serious desktop card can handle once the weights are compressed. 93 of the cards we track can run it.

The least hardware that works is a Radeon PRO V710. Its 28 GB is enough at Q3_K_M compression, giving roughly 9.2 tokens per second.

A B200 is the fastest we calculate for it: about 69.2 tokens per second, from 8,000 GB/s of memory bandwidth.

About this model

Llama Nemotron Super 49B was published by NVIDIA, in United States of America, in March 2025. It comes out of industry.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Quantitative reasoning, Code generation, Neural Architecture Search - NAS.

It is derived from Llama 3.3 70B rather than trained from scratch, which is the usual way a specialised model is produced.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all. It is published under the nvidia organisation on Hugging Face.

How fast it runs, and why

The median result is around 17.5 tokens per second; 71 cards produce text faster than most people read it.

Being dense, it reads all of itself per token, which is why the ordering by speed below follows the ordering by memory bandwidth so closely.

Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

What went into building it

Around 100,000,000,000 tokens went into training it.

Step by step

How to choose a GPU for Llama Nemotron Super 49B

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Start from the memory column

The table lists every card that can hold Llama Nemotron Super 49B — around 24.6 GB at Q3_K_M. That figure, not the card's headline performance, is what decides whether it runs.
02

Set the context length you will work at

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for Llama Nemotron Super 49B.
03

Decide how much compression you will accept

Compression is what makes Llama Nemotron Super 49B fit smaller cards, at some cost in accuracy — Q3_K_M on the smallest card that fits. A minimum quality removes the ones that go too far.
04

Rank by throughput rather than spec sheet

Ranking by tokens per second for Llama Nemotron Super 49B follows memory bandwidth, not core counts, which is why the B200 tops it at 69.2 tok/s.
05

Read the fit column last

Tight means Llama Nemotron Super 49B loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.
06

Open the card you have settled on

Following a card through to its own page shows every other model it can hold, which is the question that follows once Llama Nemotron Super 49B is settled.

Answers

Llama Nemotron Super 49B — common questions

Can I run Llama Nemotron Super 49B if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded Llama Nemotron Super 49B is rarely worth using — the nearest miss we calculate is short by 8.7 GB. Every figure here assumes the whole model is on the card.

Would two GPUs run Llama Nemotron Super 49B faster?

Capacity adds across cards; throughput does not. Since 93 of the cards we track already hold Llama Nemotron Super 49B on their own, a second card is rarely the answer here.

Why does the quantisation differ between cards for Llama Nemotron Super 49B?

A larger card holds a more accurate copy. Across the cards that run Llama Nemotron Super 49B, 5 compression levels are used; the floor control above pins it to one.

How accurate are these Llama Nemotron Super 49B speed estimates?

These are estimates with real error bars. The fastest result here, 41–111 tok/s on the B200, could reasonably land anywhere in its published range depending on which runtime you use.

What GPU do I need to run Llama Nemotron Super 49B?

The smallest card in our catalogue that holds Llama Nemotron Super 49B is the Radeon PRO V710, with 28 GB of memory. It runs the model at Q3_K_M using about 24.6 GB, and produces roughly 9.2 tokens per second. 93 cards in total can run it.

How fast is Llama Nemotron Super 49B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 69.2 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 71 of the cards that can run Llama Nemotron Super 49B clear that.

How much VRAM does Llama Nemotron Super 49B need?

About 24.6 GB at Q3_K_M compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Is Llama Nemotron Super 49B open source?

Its weights are published, so Llama Nemotron Super 49B can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does Llama Nemotron Super 49B have?

Llama Nemotron Super 49B has 49B parameters. The model is a derivative of Meta’s Llama-3.3-70B-Instruct, using Neural Architecture Search (NAS). The NAS algorithm results in non-standard and non-repetitive blocks. This includes the following: Skip attention: In some blocks, the attention is skipped entirely, or replaced with a single linear layer. Variable FFN: The expansion/compression ratio in the FFN layer is different between blocks. We utilize a block-wise distillation of the reference model, where for each block we create multiple variants providing different tradeoffs of quality vs. computational complexity, discussed in more depth below. We then search over the blocks to create a model which meets the required throughput and memory (optimized for a single H100-80GB GPU) while minimizing the quality degradation. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created Llama Nemotron Super 49B?

Llama Nemotron Super 49B was published by NVIDIA, based in United States of America, categorised as industry.

When was Llama Nemotron Super 49B released?

Llama Nemotron Super 49B was published in March 2025.

What is Llama Nemotron Super 49B used for?

Llama Nemotron Super 49B works in Language, and is recorded as handling language modeling/generation, Question answering, Quantitative reasoning, Code generation, Neural Architecture Search - NAS. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download Llama Nemotron Super 49B?

Its weights are published under the nvidia organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models