Use DeepSeek R1 and V3 as a free OpenAI-compatible endpoint in Cursor, VS Code, Cline, Continue, and any AI application.
1Star the Project on GitHub
This proxy is 100% free and open-source. Starring the project repository on GitHub is mandatory before accessing the setup.
Mandatory RequirementStep 1 of 5
Support This Free Tool
Starring on GitHub helps keep this project active, maintained against DeepSeek web updates, and free for the community.
Important Instructions:
1. Click Star on GitHub below to open the official repository in a new tab.
2. On GitHub, click the Star button near the top right of the page.
3. Come back to this tab and click I Have Starred: Continue to Step 2 to proceed.
GitHub opened in a new tab. After clicking the Star button on GitHub, click I Have Starred: Continue to Step 2 above.
CompletedGitHub Repository Starred
2Browser Tool: 1-Click Key Grabber
Add this browser bookmarklet tool to your bookmarks bar. It allows you to extract your active DeepSeek session token with a single click while on chat.deepseek.com.
While on the chat.deepseek.com tab, click the DeepSeek Key Grabber bookmark in your browser bar.
3
Key Copied to Clipboard
A popup will confirm the key is copied. Your API key is now on your clipboard, ready to use.
Paste the copied key here (Ctrl + V). It will automatically be formatted and saved for your local server and IDE.
4Download & Run Local Server
Follow these 2 simple commands in your terminal to download and start your local server:
npm install -g freeseek
Open your terminal (Command Prompt on Windows, Terminal on Mac/Linux) and paste the command above.
freeseek
Type freeseek and press Enter. (Windows alternative: download project and double-click start.bat)
Checking
Checking if server is running on http://127.0.0.1:8787...
5Connect to Your AI App
Configure your IDE or AI application with these parameters. Notice that the Server (Base URL) and API Parameter (API Key) are two completely separate settings:
Server Address
Paste this into the Base URL, API Host, or Server URL field in your application settings:
http://127.0.0.1:8787/v1
Auth Parameter
Paste this into the API Key or Bearer Token field in your application settings:
dummy
Note: Since freeseek runs locally on your machine, you can use your DeepSeek key from Step 3 OR enter dummy (the local server manages authentication automatically).
Model ID
Enter this in the Model ID or Model Name field in your application settings:
deepseek-reasoner
Use deepseek-reasoner for DeepSeek-R1 extended reasoning, or deepseek-chat for standard conversation without reasoning blocks.
Cursor Instructions
VS Code / Cline
Other AI Apps
Python SDK
1. Open Cursor Settings (gear icon).
2. Go to "Models" -> "OpenAI".
3. Turn ON "Override OpenAI Base URL" and enter:
Server (Base URL): http://127.0.0.1:8787/v1
4. In "OpenAI API Key" (API Parameter), enter:
API Parameter: dummy (or paste your key from Step 3)
5. In model list, add: deepseek-reasoner (or deepseek-chat)
In Cline / Continue / Roo Code settings:
Provider: OpenAI Compatible
Server (Base URL): http://127.0.0.1:8787/v1
API Parameter (API Key): dummy
Model Parameter: deepseek-reasoner
In NextChat, Chatbox, or LibreChat:
API Provider: OpenAI Compatible
Server (API Host / Base URL): http://127.0.0.1:8787/v1
API Parameter (API Key): dummy
Model Parameter: deepseek-reasoner
from openai import OpenAI
# Server parameter points to local freeseek proxy
# API parameter is dummy (or your DeepSeek session token)
client = OpenAI(
base_url="http://127.0.0.1:8787/v1",
api_key="dummy"
)
response = client.chat.completions.create(
model="deepseek-reasoner",
messages=[{"role": "user", "content": "Hello!"}],
stream=True
)
for chunk in response:
print(chunk.choices[0].delta.content or "", end="")