GitHub Copilot

​

Prerequisites

• A Fireworks API key
• GitHub Copilot with access to Other Models and Custom Endpoint (availability depends on your Copilot plan)

​

Setup

1

Open model settings

In Copilot Chat, click the active model name at the bottom (often Auto). In the menu, click the gear icon next to Other Models.

2

Add a custom endpoint

In Language Models, click + Add Models… in the top right, then choose Custom Endpoint.

3

Name the endpoint group

Enter Fireworks AI as the group name and press Enter.

4

Enter your API key

Paste your Fireworks API key (hidden by default) and press Enter to confirm.

5

Choose Responses API

When asked for the default request/response format, select Responses API.

6

Configure the model template

A configuration file opens. Do not change the auto-generated header at the top—only fill in the model template below it.

7

Add model details and save

Fill in your model fields, then save (Ctrl+S on Windows/Linux, Cmd+S on macOS) and close the settings modal.Example for DeepSeek V4 Pro:

Field	Value
id	accounts/fireworks/models/deepseek-v4-pro
name	DeepSeek V4 Pro
url	https://api.fireworks.ai/inference/v1
toolCalling	true
vision	false
maxInputTokens	1000000
maxOutputTokens	384000

Use the exact model id and token limits from the model page in the Model Library. Values differ per model.

8

Select your model in chat

Return to Copilot Chat, open the model picker (Auto), expand Other Models, and choose your model under Fireworks AI.

​

Related

• Claude Code — use Fireworks models with Claude Code
• Development Setup with Fireworks Docs MCP — add Fireworks docs to your coding agent