Run Meta's Llama 4 Scout, a Mixture-of-Experts model with 17B active parameters, behind an OpenAI-compatible API in your own AWS account. No dependency wrangling or GPU driver setup.
Llama 4 Scout is part of Meta's Llama 4 open model generation. It uses a Mixture-of-Experts architecture with 17 billion active parameters out of 109 billion in total, for strong text generation, chat and embeddings.
17B active parameters per token, 16 experts.
Writing, translation and question answering.
Generate and explain code.
Chat, completions, embeddings and model listing.
Test endpoints from the dashboard URL.
Prompts and outputs stay in your own account.
The AMI ships Llama 4 Scout and an OpenAI-compatible API server on Ubuntu 22.04, with GPU drivers ready.
Make sure your account has vCPU quota for g4dn instances in your region.
Subscribe on AWS Marketplace and launch the stack. Double-check the admin email; it cannot be changed later.
After 5-10 minutes, create a DNS record with the PublicIp from the stack outputs.
Open the DashboardUrl to test the endpoints, then point your OpenAI client at your server.
An AMI on AWS Marketplace, launched on a g4dn GPU instance in your own AWS account with a CloudFormation stack.
Launch on AWS | Endpoint | Purpose |
|---|---|
| /v1/chat/completions | Chat completions from a list of messages |
| /v1/completions | Completions from a prompt |
| /v1/embeddings | Embeddings for input text |
| /v1/models | List available models |
The developer guide includes a small Node.js script that checks every endpoint. Read the Llama 4 developer guide →
A Meta Llama 4 model with a Mixture-of-Experts architecture: 17 billion active parameters, 109 billion in total.
The AWS listing recommends g4dn.metal.
Yes. It serves the OpenAI chat, completions, embeddings and models endpoints.
No. It is used for the SSL certificate and cannot be changed after the stack is created.
SSH in and restart the llama service, then check the disk is not full.
The model is still loading. Wait about 5 minutes and refresh.
Back up your server data, remove the old stack and launch the new version from AWS Marketplace.
Use cases, technical highlights and the Meetrix AMI.
How Llama 4 compares with other open models.
Launch the stack, set up SSL and test the API.
We build and run self-hosted AI infrastructure every day, from model selection to GPU sizing. Tell us what you need.
Contact us