Llama Nemotron Ultra 253B TPS calculator

Open weights NVIDIA 253B 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

9 of 818 cards that can run it

Smallest card that fits

H200 NVL

141 GB · Q3_K_M · 22.1 tok/s

Fastest card

B200

30.9 tok/s · 180 GB

Which GPUs can run Llama Nemotron Ultra 253B?

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

9 cards match

Calculating

					Needs	Quantisation	Fit
30.9 tok/s 19–49 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	153.8 GB	Q4_K_M	Tight
22.1 tok/s 13–35 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	124.3 GB	Q3_K_M	Tight
22.1 tok/s 13–35 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	124.3 GB	Q3_K_M	Tight
19.5 tok/s 12–31 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	212.7 GB	Q6_K	Comfortable
16.1 tok/s 10–26 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	153.8 GB	Q4_K_M	Tight
16.1 tok/s 10–26 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	153.8 GB	Q4_K_M	Tight
15.5 tok/s 9–25 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	212.7 GB	Q6_K	Comfortable
15.5 tok/s 9–25 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	212.7 GB	Q6_K	Comfortable
11.4 tok/s 7–18 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	212.7 GB	Q6_K	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 3.1-405B

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: 253B
Training data: 603,000,000,000 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.

Training compute: 3.9 × 10²⁵ FLOP
How it was established: Operation counting
Fine-tuning compute: 1.1 × 10²⁴ 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.

Frontier model: Yes
Likely above 10²³ FLOP: Yes
Record confidence: Likely

Sources

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

Reference: Ultra is 253B distilled from Llama 3.1 405B for maximum agentic accuracy on multi-GPU data center servers.
Last updated: 21 July 2026

The extremes

The ten fastest GPUs for Llama Nemotron Ultra 253B

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 Ultra 253B

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

H200 NVL

Memory needed

124.3 GB

Fastest

30.9 tok/s

At 253B parameters, Llama Nemotron Ultra 253B is beyond what any single graphics card holds. Running it means either splitting it across several cards or renting hardware built for the job — 9 of the cards we track can hold it on their own, and all of them are datacentre parts.

The least hardware that works is a H200 NVL. Its 141 GB is enough at Q3_K_M compression, giving roughly 22.1 tokens per second.

At the other end, a B200 generates roughly 30.9 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

Where it came from

Llama Nemotron Ultra 253B was published by NVIDIA, in United States of America, in March 2025. industry is the category the publisher falls under.

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 3.1-405B rather than trained from scratch, which is the usual way a specialised model is produced.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached. It is published under the nvidia organisation on Hugging Face.

Understanding the speeds

Across every card that can run it, the middle of the range is about 16.1 tokens per second, and 9 of them clear the ten tokens per second that roughly matches reading speed.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

Training and provenance

The training run consumed about 3.9 × 10²⁵ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

The training set ran to roughly 603,000,000,000 tokens.

Step by step

How to choose a GPU for Llama Nemotron Ultra 253B

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

01

Check what it needs before anything else

Every card here has been checked against Llama Nemotron Ultra 253B — around 124.3 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
02

Decide how long your conversations run

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context Llama Nemotron Ultra 253B can slip off a card that handles short questions easily.
03

Set a quality floor

The quantisation column varies by card, because a bigger card holds a more accurate copy of Llama Nemotron Ultra 253B — Q3_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

The speed ordering for Llama Nemotron Ultra 253B is effectively an ordering by memory bandwidth, which is why the B200 tops it at 30.9 tok/s.
05

Look at the headroom, not just the fit

A tight fit runs Llama Nemotron Ultra 253B but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
06

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once Llama Nemotron Ultra 253B is settled.

Answers

Llama Nemotron Ultra 253B — common questions

What GPU do I need to run Llama Nemotron Ultra 253B?

The smallest card in our catalogue that holds Llama Nemotron Ultra 253B is the H200 NVL, with 141 GB of memory. It runs the model at Q3_K_M using about 124.3 GB, and produces roughly 22.1 tokens per second. 9 cards in total can run it.

How fast is Llama Nemotron Ultra 253B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 30.9 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 9 of the cards that can run Llama Nemotron Ultra 253B clear that.

How much VRAM does Llama Nemotron Ultra 253B need?

About 124.3 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 Ultra 253B open source?

Its weights are published, so Llama Nemotron Ultra 253B 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 Ultra 253B have?

Llama Nemotron Ultra 253B has 253B parameters. 253B "Dense decoder-only Transformer model Network Architecture: Llama-3.1-405B-Instruct, customized through Neural Architecture Search (NAS) **This model was developed based on Llama-3.1-405B-Instruct ** This model has 253B model parameters.". 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 Ultra 253B?

Llama Nemotron Ultra 253B was published by NVIDIA, based in United States of America, categorised as industry.

When was Llama Nemotron Ultra 253B released?

Llama Nemotron Ultra 253B was published in March 2025.

What is Llama Nemotron Ultra 253B used for?

Llama Nemotron Ultra 253B works in Language, and is recorded as handling language modeling/generation, Question answering, Quantitative reasoning, Code generation, Neural Architecture Search - NAS. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

Where can I download Llama Nemotron Ultra 253B?

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.

How much compute was used to train Llama Nemotron Ultra 253B?

Around 3.9 × 10²⁵ FLOP. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Can I run Llama Nemotron Ultra 253B 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 Ultra 253B is rarely worth using — the nearest miss we calculate is short by 38.6 GB. Every figure here assumes the whole model is on the card.

Would two GPUs run Llama Nemotron Ultra 253B faster?

Capacity adds across cards; throughput does not. Since 9 of the cards we track already hold Llama Nemotron Ultra 253B on their own, a second card is rarely the answer here.

Why does the quantisation differ between cards for Llama Nemotron Ultra 253B?

Because capacity varies, so does how hard Llama Nemotron Ultra 253B has to be squeezed — 3 distinct levels appear in the table above. Set a minimum quality to compare at one.

How accurate are these Llama Nemotron Ultra 253B speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 19–49 tok/s on the B200 rather than a single number.

Source

Original publication

Record last updated 21 July 2026

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