llmrun
NVIDIALlama 3DeciLMForCausalLM

Llama 3 1 Nemotron Ultra 253B V1 — Hardware Requirements & GPU Compatibility

Chat

Llama 3 1 Nemotron Ultra 253B V1 is a 253.4B-parameter open language model from NVIDIA in the Llama 3 family. It supports a context window of up to 131,072 tokens. At BF16 it needs about 557.48 GB of VRAM — see which GPUs and Macs can run it below.

253.4B params · dense131K contextOther5.0K downloads352 likes

Llama 3 1 Nemotron Ultra 253B V1 runs on a multi-GPU server

BF16 needs ~564.2 GB · about 8× H100 SXM 80 GB

Other

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Llama 3 1 Nemotron Ultra 253B V1

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Recommended

vLLM on a multi-GPU server
tensor + pipeline parallel · needs vLLM on every node
$vllm serve nvidia/Llama-3_1-Nemotron-Ultra-253B-v1 --max-model-len 16384 --tensor-parallel-size 8

Size the nodes from the table above

Download: Weights on Hugging Face

install, other ways

Other ways

each opens step-by-step

Run in cloud

Rent 8× H100 SXM node · BF16

Estimated hourly price
~$17.69/hr
est., whole server
GPUs
8 GPUs
640 GB of VRAM
Deployon Vast.aiRent H100 SXM$2.21/hr per GPU (opens in new tab)
Compare GPUs →

How Much VRAM Does Llama 3 1 Nemotron Ultra 253B V1 Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KRuns on
557.5 GB572 GB16-bit · bfloat — Brain floating point 16 — preferred for trainingRuns on 8× H100 SXM 80 GB
est.calculated, no published file yet

Selected BF16

Brain floating point 16

564.2GBat 16K context

Weights
556.2 GB
KV cache
7.7 GB
Overhead
0.3 GB

Server-class: ~564.2 GB at 16K context · about 8× H100 SXM 80 GB

Which Hardware Can Run Llama 3 1 Nemotron Ultra 253B V1?

BF16 · 557.5 GB

Llama 3 1 Nemotron Ultra 253B V1 (BF16) requires 557.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 725+ GB is recommended. No consumer GPU has enough memory.

Needs server-class hardware

Llama 3 1 Nemotron Ultra 253B V1 runs on a multi-GPU server · BF16 needs ~564.2 GB

Hardware that runs this model
HardwareQuantMemoryContextHardware page
8× H100 SXM 80 GB · 640 GBBF16564.2 GB32K ctxH100 SXM →BF16 · 564.2 GB · 32K ctx
5× H200 SXM 141 GB · 705 GBBF16564.2 GB128K ctxH200 SXM →BF16 · 564.2 GB · 128K ctx
4× B200 192 GB · 768 GBBF16564.2 GB128K ctxB200 →BF16 · 564.2 GB · 128K ctx

Also: NVIDIA DGX A100 640GB · NVIDIA DGX H100

≤ 640 GB tier →

By VRAM tier · typical GPUs and Macs at BF16

Consumer GPUs (up to 32 GB) and Macs up to 512 GB: too small for this model (needs ≥ 564.2 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →

How good is Llama 3 1 Nemotron Ultra 253B V1?

llmrun Score blends public benchmark boards into one number; 100 = the average open model.

Not enough data yet for an llmrun Score.

A Score needs results on enough independent boards. How the Composite Score is built

Benchmarks

This is the single board showing results for this model; badges show who ran the test.

Llama 3 1 Nemotron Ultra 253B V1 benchmark results

Alternatives to Llama 3 1 Nemotron Ultra 253B V1

Llama 3 family · 64 models →

Similar models

3
ModelScoreVRAMWhy it is similar
Inkling SmallThinking Machines119.6162.7 GBsame size
Qwen3 VL 235B A22B InstructAlibaba—143.3 GBsame size
DeepSeek V2.5 1210DeepSeek—142.9 GBsame size

Specifications

Publisher
NVIDIA
Family
Llama 3
Parameters
253.4B
Architecture
DeciLMForCausalLM
Context Length
131,072 tokens
Vocabulary Size
128,256
Release Date
2025-04-07
License
Other

Resources

Frequently Asked Questions

How much VRAM does Llama 3 1 Nemotron Ultra 253B V1 need?

Llama 3 1 Nemotron Ultra 253B V1 requires 557.5 GB of VRAM at BF16.

VRAM = Weights + KV Cache + Overhead

Weights = 253.4B × 16 bits ÷ 8 = 506.8 GB

KV Cache + Overhead ≈ 50.7 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

VRAM usage by quantization

557.5 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run Llama 3 1 Nemotron Ultra 253B V1?

No — Llama 3 1 Nemotron Ultra 253B V1 requires at least 557.5 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

Can I run Llama 3 1 Nemotron Ultra 253B V1 on a Mac?

Llama 3 1 Nemotron Ultra 253B V1 requires at least 557.5 GB at BF16. None of the 146 Macs we list has that much usable memory (the largest has 508.5 GB after macOS overhead).

Can I run Llama 3 1 Nemotron Ultra 253B V1 locally?

Yes — Llama 3 1 Nemotron Ultra 253B V1 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

What's the download size of Llama 3 1 Nemotron Ultra 253B V1?

At BF16, the download is about 506.80 GB.

Which GPUs can run Llama 3 1 Nemotron Ultra 253B V1?

No single consumer GPU has enough VRAM to run Llama 3 1 Nemotron Ultra 253B V1 at BF16 (557.5 GB). Multi-GPU or professional hardware is required.

Which devices can run Llama 3 1 Nemotron Ultra 253B V1?

5 GB).