LLM Hosting Pro
← Llama family

Llama 3.1 8B Instruct

A small, fast Llama model that runs on a single GPU.

Overview

Llama 3.1 8B Instruct is a small, fast model from Meta's Llama 3.1 family. It runs on a single GPU, supports a 128K-token context window and is widely used for chat, classification and as a base for fine-tuning. It's one of the cheapest open models to host.

Model specifications
WeightsOpen
Parameters8B
Context window128K tokens
LicenceLlama Community
Commercial useWith conditions

LLM hosts that offer Llama 3.1 8B Instruct

All hosts →
LLM hostHosting planHosting typePricing
Amazon AWSAmazon BedrockServerless APIFrom US$0.04Pay per token, Reserved or committed capacity
Amazon AWSAmazon SageMakerDedicated endpointFrom US$0.736Per GPU-hour, Reserved or committed capacity
Amazon AWSAmazon EC2GPU cloudFrom US$0.526Per GPU-hour, Reserved or committed capacity
Google CloudGoogle Kubernetes Engine (GKE)GPU cloudFrom US$0.35Per GPU-hour, Reserved or committed capacity
Google CloudCompute EngineGPU cloudFrom US$0.35Per GPU-hour, Reserved or committed capacity
Microsoft AzureAzure Machine LearningDedicated endpointFrom US$0.526Per GPU-hour, Reserved or committed capacity
Microsoft AzureAzure Kubernetes Service (AKS)GPU cloudFrom US$0.526Per GPU-hour, Reserved or committed capacity
Microsoft AzureAzure GPU Virtual MachinesGPU cloudFrom US$0.526Per GPU-hour, Reserved or committed capacity