Free OpenAI API Access via GitHub Copilot + LiteLLM

instructions.md

Free OpenAI API Access via GitHub Copilot + LiteLLM

If you have a .edu email, you can get GitHub Copilot Pro for free through GitHub Education. Here's the sign-up guide.

Copilot Pro gets you access to models from OpenAI, Claude, Gemini, Grok and more. The VS Code integration is nice but what's really useful is that you can expose all of this as a local OpenAI-compatible API endpoint, meaning any tool or script that speaks the OpenAI API just works with it, for free.

The trick is LiteLLM, which acts as a proxy between your code and whatever model backend you want. All you need is one config file.

What you need

• GitHub Education access with Copilot Pro enabled
• Python 3.11. Avoid 3.14, it breaks some of LiteLLM's proxy dependencies
• VS Code to get slugs of the models you want to use.

Step 1: Install LiteLLM

Spin up a venv and install the proxy extras:

python3.11 -m venv .venv
source .venv/bin/activate        # Windows: .venv\Scripts\activate
pip install 'litellm[proxy]'

Step 2: Create a config file

Paste this into a file called litellm-config.yaml:

model_list:
  - model_name: gpt-4.1
    litellm_params:
      model: github_copilot/gpt-4.1
  - model_name: gpt-5.1-codex
    litellm_params:
      model: github_copilot/gpt-5.1-codex
  - model_name: claude-sonnet-4.5
    litellm_params:
      model: github_copilot/claude-sonnet-4.5
  - model_name: gemini-3-pro
    litellm_params:
      model: github_copilot/gemini-3-pro-preview

model_name is the alias you'll use in your API calls. litellm_params.model is the actual Copilot model slug, it always starts with github_copilot/. You can add as many entries as you want:

model_name (your alias)	litellm_params.model (Copilot slug)
gpt-4.1	github_copilot/gpt-4.1
claude-sonnet-4.6	github_copilot/claude-sonnet-4.6
gemini-3-pro	github_copilot/gemini-3-pro-preview
gpt-5.1-codex	github_copilot/gpt-5.1-codex

Finding the slug for a model: Open VS Code, go to the Copilot Chat panel, and click the model dropdown at the top. The last option is Manage Models. Click it to see all available models. Hover over any model name and you'll see its slug. Prepend github_copilot/ to it and you're good.

No API keys to set. LiteLLM handles Copilot auth on its own (more on that below).

Step 3: Start the proxy

litellm --config litellm-config.yaml

The first time you run this, LiteLLM will prompt you to authenticate with GitHub. Just go through the OAuth flow — it only happens once. After that, your token is cached.

Once you're authenticated, you should see:

LiteLLM: Proxy initialized with Config, Set models:
    gpt-4.1
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:4000 (Press CTRL+C to quit)

---

Step 4 — Test it

Swap the model below to one of the aliases you defined in the YAML file.

curl http://localhost:4000/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4.1",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

If it's working, you'll get back something like:

{
  "id": "xxx",
  "model": "gpt-4.1",
  "object": "chat.completion",
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help you today? 😊"
      }
    }
  ],
  "usage": {
    "prompt_tokens": 8,
    "completion_tokens": 11,
    "total_tokens": 19
  }
}

And you've got an OpenAI-compatible API server running on your machine!

Using it in your code

Since the proxy speaks the OpenAI protocol, you just point the SDK at localhost:4000:

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:4000",
    api_key="anything",   # no real key needed
)

response = client.chat.completions.create(
    model="gpt-4.1",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

Or you can use LiteLLM's SDK directly without starting the proxy server. More details on that here

Heads up on rate limits: Copilot Pro rate limits still apply. The proxy doesn't bypass them.

Issues I ran into

Problem	Fix
litellm command not found	You probably forgot to activate the venv: source .venv/bin/activate
Weird Python dependency errors	Switch to Python 3.11 and avoid 3.14