Llama-3.1-Nemotron-Ultra-253B-v1

NVIDIA

Container

NVIDIA

Llama-3.1-Nemotron-Ultra-253B-v1

This container houses the Llama-3.1-Nemotron-Ultra-253B-v1, a reasoning model offering a great tradeoff between accuracy and efficiency. Post-trained for chat, RAG, and tool calling, it supports a 128K context and fits on a single 8xH100 node.

NVIDIA NIM

NIM MetadataTo verify your workspace, run the Verify CLI locally and compare the generated hash with the NGC-published hash shown here. Learn more in the NVIDIA Documentation.

			Profile	Engine		Actions
H200 NVL	233b:10de	bf16	latency	tensorrt_llm	8	Copy NIM CLI Verification
H100 80GB HBM3	2330:10de	bf16	throughput	tensorrt_llm	8	Copy NIM CLI Verification
H200 NVL	233b:10de	fp8	throughput	tensorrt_llm	8	Copy NIM CLI Verification
H200	2335:10de	fp8	latency	tensorrt_llm	8	Copy NIM CLI Verification
H100 80GB HBM3	2330:10de	fp8	latency	tensorrt_llm	8	Copy NIM CLI Verification
H200 NVL	233b:10de	fp8	latency	tensorrt_llm	8	Copy NIM CLI Verification
H200 NVL	233b:10de	bf16	throughput	tensorrt_llm	8	Copy NIM CLI Verification
B200	2901:10de	fp8	throughput	tensorrt_llm	8	Copy NIM CLI Verification
H100 NVL	2321:10de	bf16	latency	tensorrt_llm	8	Copy NIM CLI Verification
H100 NVL	2321:10de	fp8	throughput	tensorrt_llm	8	Copy NIM CLI Verification