@font-face {
  font-family: "Saans-Light";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-Light.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-Light.woff") format("woff");
  font-weight: 300;
  font-style: normal;
  font-display: swap;
}
@font-face {
  font-family: "Saans-Regular";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-Regular.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-Regular.woff")
      format("woff");
  font-weight: 400;
  font-style: normal;
  font-display: swap;
}
@font-face {
  font-family: "Saans-Medium";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-Medium.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-Medium.woff")
      format("woff");
  font-weight: 500;
  font-style: normal;
  font-display: swap;
}
@font-face {
  font-family: "Saans-SemiBold";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-SemiBold.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-SemiBold.woff")
      format("woff");
  font-weight: 600;
  font-style: normal;
  font-display: swap;
}
@font-face {
  font-family: "Saans-Bold";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-Bold.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-Bold.woff") format("woff");
  font-weight: 700;
  font-style: normal;
  font-display: swap;
}
@font-face {
  font-family: "Saans-Heavy";
  src:
    url("https://docs-assets.friendli.ai/fonts/Saans-Heavy.woff2")
      format("woff2"),
    url("https://docs-assets.friendli.ai/fonts/Saans-Heavy.woff") format("woff");
  font-weight: 800;
  font-style: normal;
  font-display: swap;
}

body {
  font-family: "Saans-Regular", sans-serif;
  letter-spacing: 0.01em;
}

.tracking-tighter {
  letter-spacing: -0.04em;
}
.tracking-tight {
  letter-spacing: 0.01em;
}
.tracking-normal {
  letter-spacing: 0.01em;
}
.tracking-wide {
  letter-spacing: 0.01em;
}
.tracking-wider {
  letter-spacing: 0.05em;
}
.tracking-widest {
  letter-spacing: 0.1em;
}
.prose
  :where(h1, h2, h3):not(:where([class~="not-prose"], [class~="not-prose"] *)) {
  letter-spacing: 0;
}

.card > div > div:last-child > h2 {
  font-size: 1.25rem;
}
.card > div > div:last-child > div {
  margin-top: 1rem;
}

.font-medium:not(.font-mono):not(.font-mono *) {
  font-family: "Saans-Medium", sans-serif;
}

h1:not(.font-mono):not(.font-mono *),
h2:not(.font-mono):not(.font-mono *),
h3:not(.font-mono):not(.font-mono *),
h4:not(.font-mono):not(.font-mono *),
h5:not(.font-mono):not(.font-mono *),
strong:not(.font-mono):not(.font-mono *),
.font-semibold:not(.font-mono):not(.font-mono *),
.prose
  :where(a):not(:where([class~="not-prose"], [class~="not-prose"] *)):not(
    .font-mono
  ):not(.font-mono *) {
  font-family: "Saans-SemiBold", sans-serif;
}

#header > div:not(:first-child) {
  display: none;
}

#topbar-cta-button > .font-medium {
  font-family: "Saans-Medium", sans-serif;
}

#navbar-transition {
  --tw-backdrop-blur: blur(0px);
}

#banner {
  background-color: #2e70ff;
  font-size: 16px;
  box-shadow: none;
}

#banner > div {
  margin-top: 0.625rem;
  margin-bottom: 0.625rem;

  a {
    color: #ffffff;
    font-weight: 600;
    text-decoration: underline;
    padding-left: 0.25rem;
  }
}

.border-gray-100 {
  border-color: #d9e2ec;
}

:root {
  --gray-300: 159 179 200;
}

.dark\:bg-codeblock div[role="tablist"] button[role="tab"] {
  /* prettier-ignore */
  font-family:
    var(--font-jetbrains-mono), ui-monospace, SFMono-Regular, Menlo, Monaco,
    Consolas, "Liberation Mono", "Courier New", monospace;
}

.rounded-border-box {
  padding: 28px 44px;
  border: 1px solid #d9e3ee;
  border-radius: 8px;
  overflow: hidden;

  img {
    margin: 0 auto;
    border: 1px solid #e7eef4;
    border-radius: 8px;
  }

  p {
    margin: 12px auto -12px;
    text-align: center;
    word-break: break-all;
  }
}

.dark {
  .rounded-border-box {
    border: 1px solid #343842;

    img {
      border: 1px solid #212832;
    }
  }
}


import * as mdx from "eslint-plugin-mdx";

export default [
  {
    files: ["**/*.mdx"],
    ...mdx.flat,
    processor: mdx.createRemarkProcessor({
      lintCodeBlocks: true,
      languageMapper: {},
    }),
    rules: {
      "no-unused-expressions": "off",
    },
  },
  {
    files: ["**/*.mdx"],
    ...mdx.flatCodeBlocks,
    rules: {
      ...mdx.flatCodeBlocks.rules,
      "no-var": "error",
      "prefer-const": "error",
    },
  },
];


Supported Instance Types

Supported Model Sizes

Friendli Dedicated Endpoints pricing detail page.

Pricing

Running real inference at scale? Apply for our limited $10K credit program — [Find out more](https://friendli.ai/promotion10k)

Run Large language models (LLMs) & multimodal models faster, cheaper, and easier. FriendliAI slashes GPU costs with high-throughput inference—no infra ops required.

Friendli Docs

Get started with FriendliAI products and explore APIs.

Overview

Friendli Documentation

Friendli Suite empowers you to explore generative AI with three solutions: Serverless Endpoints for quick access to open-source models, Dedicated Endpoints for deploying custom models on dedicated GPUs, and Containers for secure, on-premise control. Powered by the optimized Friendli Engine, each option ensures fast, cost-efficient AI serving for text, code, and image generation.

Introduction

Unleash the Power of Generative AI with Friendli Suite: Your End-to-End Solution

Learn how to manage credentials in Friendli Suite, including using personal access tokens for authentication and authorization.

Personal Access Tokens

Dive into the characteristics of popular Text Generation Models (TGMs) available on Friendli.

Text Generation

Text Generation Models

Dive into the characteristics of popular Image Generation Models available on Friendli Dedicated Endpoints.

Image Generation

Visualizing Ideas with Friendli: A Guide to Image Generation

Guide to using Friendli's Vision feature for image analysis. Covers usage via Playground and API (URL & Base64 examples).

Vision

Vision: Image understanding with Friendli

Guide to using Friendli's Audio and Speech feature for audio analysis and transcription. Covers usage via Playground and API (URL & Base64 examples).

Audio and Speech

Audio and Speech: Converting Audio to Text

Learn how to do OpenAI compatible function calling on Friendli Serverless Endpoints.

Function Calling

Generate structured outputs using FriendliAI's Structured Outputs feature.

Structured Outputs

Friendli Serverless Endpoints is compatible with the OpenAI API standard through the Python API Libraries and the Node API Libraries. Friendli Dedicated Endpoints and Friendli Container are also OpenAI API compatible.

OpenAI Compatibility

Build AI agents with Friendli Serverless Endpoints using tool calling for dynamic, real-time interactions with LLMs.

Tool calling with Serverless Endpoints

Hands-on tutorial for launching and deploying LLMs using Friendli Dedicated Endpoints with Weights & Biases artifacts through webhook automation.

Deploy from W&B Registry with Webhook

Build an AI agent with LangChain and Friendli Serverless Endpoints, integrating tool calling for dynamic and efficient responses.

Build an agent with LangChain

Build and deploy smart AI agents with Friendli Serverless Endpoints and Gradio in under 50 lines.

Build an agent with Gradio

Go Playground with Next.js

Chat docs with LangChain

Chat docs with MongoDB

RAG app with LlamaIndex

Interact with Friendli AI services using the official Python SDK for seamless integration with your applications.

Python SDK

Friendli Python SDK

Effortlessly integrate FriendliAI models into your projects with support for popular SDKs and frameworks.

Friendli Integrations

Easily integrate FriendliAI models with the Vercel AI SDK, supporting serverless, dedicated, and fine-tuned endpoints.

Vercel AI SDK

Easily integrate large language models with the LlamaIndex SDK, featuring FriendliAI for seamless interaction.

LlamaIndex

LiteLLM SDK supports all FriendliAI models, offering easy integration with serverless, dedicated, and fine-tuned endpoints.

LiteLLM

OpenAPI reference of Friendli Suite API. You can interact with the API through HTTP requests from any language.

Friendli Suite API Reference

OpenAPI reference of Friendli Dedicated Endpoints API.

Dedicated overview

OpenAPI reference of Friendli Serverless Endpoints API.

Serverless overview

Given a list of messages forming a conversation, the model generates a response.

Chat completions

Serverless chat completions

Given a list of messages forming a conversation, the model generates a response. Additionally, the model can utilize built-in tools for tool calls, enhancing its capability to provide more comprehensive and actionable responses.

Tool assisted chat completions

Serverless tool assisted chat completions

Generate text based on the given text prompt.

Completions

Serverless completions

By giving a text input, generate a tokenized output of token IDs.

Tokenization

Serverless tokenization

By giving a list of tokens, generate a detokenized output text string.

Detokenization

Serverless detokenization

Represents a streamed chunk of a chat completions response returned by model, based on the provided input.

Chat completions chunk object

Serverless chat completions chunk object

Represents a streamed chunk of a tool assisted chat completions response returned by model, based on the provided input.

Tool assisted chat completions chunk object

Serverless tool assisted chat completions chunk object

Represents a streamed chunk of a completions response returned by model, based on the provided input.

Completions chunk object

Serverless completions chunk object

OpenAPI reference of Friendli Container API.

Container overview

Container chat completions

Container completions

Container tokenization

Container detokenization

Given a description, the model generates image.

Image generations

Container image generations

Container chat completions chunk object

Container completions chunk object

OpenAPI reference of Friendli Dataset API.

Dataset overview

List datasets

List versions

List splits

List samples

Get dataset info

Get version info

Get split info

Create a new dataset

Create a new version

Create a new split

Add samples

Delete samples

Update samples

Delete a version

Delete a dataset

Delete a split

File overview

Get file info

Get file download URL

Initiate file upload

Complete file upload

Friendli Dedicated Endpoints gives you the reins to explore the full potential of your custom generative AI models on the hardware of your choice, whether you're crafting innovative eloquent texts, generating stunning images, or even more.

Introducing Friendli Dedicated Endpoints

Learn how to get started with Friendli Dedicated Endpoints in this step-by-step guide. Create an account, select your project, choose a model you wish to serve, deploy your endpoint, and seamlessly generate text, code, and more with ease.

QuickStart

QuickStart: Friendli Dedicated Endpoints

Friendli Dedicated Endpoints projects are a basic working unit for your team.

Projects

Within your Friendli Dedicated Endpoints projects you can prepare and manage the models that you wish to deploy. You may upload your models within your project to deploy them directly on your endpoints. Alternatively, you may manage them on the HuggingFace repository or Weights & Biases artifacts, as our endpoints can load models from your project, HuggingFace repositories, and Weights & Biases artifacts.

Models

Endpoints are the actual deployments of your models on your specified GPU resource.

Endpoints

Effortlessly fine-tune your model with Friendli Dedicated Endpoints, which leverages the Parameter-Efficient Fine-Tuning (PEFT) method to reduce training costs while preserving model quality, similar to full-parameter fine-tuning.

Fine-tuning

Hands-on tutorial for launching and deploying LLMs using Friendli Dedicated Endpoints with Weights & Biases artifacts.

Tutorial | W&B Models

Deploy with W&B Models

Hands-on tutorial for launching and deploying LLMs using Friendli Dedicated Endpoints with Hugging Face models.

Tutorial | Hugging Face Models

Deploy with Hugging Face Models

While following through our tutorials, you might have had questions regarding the details of the requirements and specifications. We have listed out the frequently asked questions and as a separate document.

FAQs and Troubleshooting

Frequently Asked Questions and Troubleshooting

While Friendli Serverless Endpoints and Dedicated Endpoints offer convenient cloud-based solutions, some users crave even more control and flexibility. For those pioneers, Friendli Container is the answer.

Introducing Friendli Container

Learn how to get started with Friendli Container in this step-by-step guide. Activate your free trial, access to the Container registry, perpare you container secret, run your Friendli Container, and monitor using Granfana.

QuickStart: Friendli Container Trial

Friendli Container enables you to effortlessly deploy your generative AI model on your own machine. This tutorial will guide you through the process of running a Friendli Container.

Running Friendli Container

The Friendli Engine supports CUDA-enabled NVIDIA GPUs, which means it relies on a specific version of CUDA and necessitates proper CUDA compute compatibilities.

CUDA Compatibility

Tutorial for serving quantized model with Friendli Engine. Friendli Engine supports FP8, IN8, and AWQ-ed model checkpoints.

Serving Quantized Models

The Friendli Engine introduces an innovative approach to this challenge through Multi-LoRA (Low-Rank Adaptation) serving, a method that allows for the simultaneous serving of multiple LLMs, optimized for specific tasks without the need for extensive retraining.

Serving Multi-LoRA Models

Explore the steps to serve Mixture of Experts (MoE) models such as Mixtral 8x7B using Friendli Container.

Serving MoE Models

For specialized cases like MoE or quantized models, optimizing the execution policy in Friendli Engine can boost inference performance by 1.5x to 2x, improving throughput and reducing latency.

Optimizing Inference with Policy Search

Create a real-time inference endpoint in Amazon SageMaker with Friendli Container backend. By utilizing Friendli Container in your SageMaker pipeline, you'll benefit from the Friendli Engine's speed and resource efficiency.

Running Friendli Container on SageMaker

Run gRPC inference server with Friendli Container and interact with it through friendli-client SDK.

Inference with gRPC

Observability is an integral part of DevOps. To support this, Friendli Container exports internal metrics in a Prometheus text format.

Monitoring

Observability for Friendli Container

Guide for Friendli Serverless Endpoints, allowing you to seamlessly integrate state-of-the-art AI models into your workflows, regardless of your technical expertise.

Introducing Friendli Serverless Endpoints

Learn how to get started with Friendli Serverless Endpoints in this step-by-step guide. Create an account, choose from powerful AI models like Llama 3.1, and seamlessly generate text, code, and more with ease.

QuickStart: Friendli Serverless Endpoints

Dive into the characteristics of six popular Text Generation Models (TGMs) available on Friendli Serverless Endpoints.

Tool Assisted API enhances a model's capabilities by integrating tools that extend its functionality beyond simple conversational interactions. By using this API, the model becomes more dynamic, providing more comprehensive and actionable responses. Currently, Friendli Serverless Endpoints supports a variety of built-in tools specifically designed for Chat Completion tasks.

Tool Assisted API

Tool Assisted API (Beta)

Friendli integrates with LangChain, LiteLLM, LlamaIndex, and MongoDB to streamline GenAI application deployment. LangChain and LlamaIndex enable tool calling AI agents and Retrieval-Augmented Generation (RAG), while MongoDB provides memory via vector databases, and LiteLLM boosts performance through load balancing.

Integrations

Stay tuned for detailed guides on how to perform tasks like Retrieval-Augmented Generation (RAG), Conditional Image Generation, Fine-tuning Custom Models.

Applications

Advanced Applications on Friendli Serverless Endpoints (Coming Soon!)

Understand the rate limits for Friendli Serverless Endpoints, including Requests per Minute (RPM) and Tokens per Minute (TPM), to ensure efficient usage of resources and balanced performance when interacting with AI models.

Rate Limits

Friendli Serverless Endpoints offer a range of models tailored to various tasks.

Integrate FriendliAI with OpenAI Python SDK for chat, streaming, and more.

Python

OpenAI Python SDK

Easily integrate FriendliAI with the OpenAI Node.js SDK.

Node.js

OpenAI Node.js SDK

Utilize the LangChain Python SDK with FriendliAI for easy integration and advanced tool calling in your applications.

LangChain Python SDK

Utilize the LangChain Node.js SDK with FriendliAI for seamless integration and enhanced tool calling capabilities in your applications.

Langchain Node.js SDK

Utilize the Weaviate to build applications with less hallucination open-source vector database.

GPU Type	Basic	Enterprise
H200 141GB	$5.9 / hour	Contact sales
H100 80GB	$4.9 / hour	Contact sales
A100 80GB	$2.9 / hour	Contact sales

Fine-tuning	Model	Basic	Enterprise
	Models ≤ 16B	$0.50 / 1M tokens	Contact sales
	Models 16.1B - 72B	$3.00 / 1M tokens	Contact sales

Get Started

Core Concepts

Products

Tutorials

Pricing

Supported Instance Types

Supported Model Sizes

Get Started

Core Concepts

Products

Tutorials

​Supported Instance Types

​Supported Model Sizes

Supported Instance Types

Supported Model Sizes