01Spike · v1.0 · stable

YourChatGPTandGemini,nowanAPI.

Use the AI subscription you already pay for in your code.No new keys. No new bills. No vendor to onboard.

Gateway · livePOST /v1/chat/completions
Gemini:6969ChatGPT:5005Unified Proxy:8000Public Tunneloff
Auto-routes by model nameOne-click public URL
gpt-4ogemini-3-proo1gpt-4-turbogemini-2.0-flasho3-minigpt-4o-minigemini-3.1-flasho1-progpt-3.5-turbogemini-3-flashgpt-4o1-minigemini-3.1-progpt-4ogemini-3-proo1gpt-4-turbogemini-2.0-flasho3-minigpt-4o-minigemini-3.1-flasho1-progpt-3.5-turbogemini-3-flashgpt-4o1-minigemini-3.1-progpt-4ogemini-3-proo1gpt-4-turbogemini-2.0-flasho3-minigpt-4o-minigemini-3.1-flasho1-progpt-3.5-turbogemini-3-flashgpt-4o1-minigemini-3.1-progpt-4ogemini-3-proo1gpt-4-turbogemini-2.0-flasho3-minigpt-4o-minigemini-3.1-flasho1-progpt-3.5-turbogemini-3-flashgpt-4o1-minigemini-3.1-pro
03The thesis

The AI you already use,
in the shape of an API.

We don't think you should pay for AI twice. So we built the wire that connects what you already have to the code you're already writing.

Open SourceMITMade for the AI community
04Everything it does

Built for every workflow. Polished desktop app.

ChatGPT Plus
Gemini Adv.
Spike
01Your subscription

Use the AI you already pay for.

Connect ChatGPT Plus or Gemini Advanced via your browser session. No new API keys, no separate billing.

Spike Chat
What exactly do you do?
I turn your ChatGPT and Gemini into a real API — no new keys, no extra cost.
Can I use it with my existing code?
Yes. Point any OpenAI-compatible client at localhost:8000 and you're done.
02Built-in chat

Test inside the app.

Switch between local and public endpoints, pick any model, and verify the full pipeline before writing a single line of code.

localhost:8000
spike.ngrok.app
03Public access

Localhost to the internet, in one click.

Toggle the switch. You'll have a stable public HTTPS URL pointing back to the AI on your machine — same address every session.

my-project
src
main.py
routes.py
spike
s
gemini_server.py
requirements.txt
package.json
04Standalone setup

Drop it in your project.

Generate a self-contained Gemini server with your tokens baked in. Four files. One command. Runs anywhere Python runs.

Save History
Use for model training
Temporary Mode Active
05Privacy first

Temporary chat by default.

Every request uses Gemini's temporary mode out of the box — nothing saved to your history unless you turn it on.

OpenAI Base URL
http://localhost:8000/v1
API Key
spike
Connected to any client
06OpenAI-compatible

Drop in any client.

Works with the OpenAI SDK, LangChain, LlamaIndex, and anything that speaks chat-completions. Change one URL, nothing else.

04For developers

Three lines of code. Any model.

Same endpoint, same JSON shape, every model. Change the model name to switch providers — Spike routes automatically.

Model — click to update example

Works with the OpenAI SDK, LangChain, LlamaIndex, and any OpenAI-format client.

import requests

response = requests.post(
    'http://localhost:8000/v1/chat/completions',
    headers={'Authorization': 'Bearer spike'},
    json={
        "model": "gpt-4o",
        "messages": [
            {"role": "user", "content": "Hello"}
        ]
    }
)

print(response.json()['choices'][0]['message']['content'])
05Get started

Two products. Pick what you need.

Both run as standalone Windows apps. No Python or Node to install.

Spikev1.0.0 · stable

Full desktop app

ChatGPT, Gemini, Unified Proxy, public tunnel, built-in chat. Everything in one Electron app.

  • Multi-provider routing
  • One-click public tunnel
  • Standalone setup wizard
  • Live chat with endpoint switcher
Download SpikeWindows · free
Spike Litev1.0.0 · stable

Just Gemini,
in your tray

Single-binary OpenAI-compatible Gemini gateway. Runs from the system tray with a web dashboard for tokens.

  • OpenAI-compatible API on :6969
  • System tray + web dashboard
  • Privacy mode by default
  • Lightweight, runs in the background
Download Spike LiteWindows · free
After install

Wire up your tokens in two minutes.

1

ChatGPT

Visit chatgpt.com/api/auth/session in your browser, copy the accessToken value, paste it into Spike.

2

Gemini

Open DevTools on gemini.google.com. Copy the __Secure-1PSID and __Secure-1PSIDTS cookies into Spike.

Stop paying twice for the same AI.

Use the ChatGPT or Gemini account you already have. Free forever, MIT licensed, runs entirely on your machine.