See how much time and money you could save with

See how MAX stacks up against the alternatives across 13 popular models and 3 instances.

BERT

BERT Large Uncased Seqlen 256

CLIP-ViT

CLIP-ViT Large Patch14

GPT

GPT-2 Small Seqlen

Llama

Llama 2/3

Mistral

Mistral 7b

Replit

Replit 3b

RoBERTa

RoBERTa Base Seqlen 128

Stable diffusion

Stable diffusion XL

Starcoder

Starcoder-7b

WavLM

WavLM Large

AMD

c5.4xlarge

ARM

c5.4xlarge

Intel

c5.4xlarge

Select a model & instance

2.7x

(Latency)

Faster on average vs PyTorch when running

Replit 3B on Intel c5.4xlarge

Try MAX now and you can benchmark performance for yourself in as little as 5 min.

Benchmark Locally