NVIDIA NIM: LLAMA-4 (Maverick) Image for performance Benchmarking on Nvidia H200 GPUs

AI & Data Science NVIDIA NIM Models
, , ,
1

Hi

I’m preparing to carry out a performance benchmark of LLAMA-4 Maverick on H200 Hardware.

I am using the process described in the NVIDIA documentation here which uses NIMs:

While the example is based on “meta/llama-3.1-8b-instruct” I require Llama-4 Maverick image, however I don’t find a llama-4 image in the Nvidia catalogue NGC (or anywhere else).

Questions:

  • Is this image available as a NIM?
  • If not, when will it be available?
  • Is there an alternative LLAMA-4 Maverick image I could use to carry out a basic non-official performance test on H200 GPUs?

Thanks in advance for any help!

2

Hi,

Please try the instructions in document link below. It has instructions to run dedicated Llama4 image on H200 GPU

https://docs.nvidia.com/nim/vision-language-models/1.4.0/examples/llama-4-maverick-17b-128e-instruct/api.html

Thank you.