If you’ve already explored AI tools like Windsurf, you might’ve noticed some recent changes, especially around pricing. Windsurf GPT-4.1 just dropped its free tier, and users of other tools are saying how expensive these options really are…
But one savvy Redditor dropped a gem 💎 that you should know about: you can get $300 in free credits through Google Cloud Platform (GCP) and use it with Vertex AI.
That means you can run Gemini 2.5 Pro at scale (for free…for now). Whether you're using any IDE like Cursor, or tools like (ahem) Kilo Code, this trick works.
😏 Wait… but doesn’t Kilo already give me $20 in free credits?
Yes. Yes, we do.
But hey—we're not here to stand between you and a free $300. If you want to use your GCP credits instead of our free credits, we’re not mad. No one likes leaving money on the table.
In fact, here’s how to do it, using instructions adapted from this Reddit thread.
How to Get $300 in GCP Credits for Vertex AI
Create or log into your Google Cloud account:
https://console.cloud.google.comCreate a new project:
Call it whatever you want.Enable billing:
You’ll get $300 in credits automatically as a new user (valid for 90 days).Enable the Vertex AI API:
Enable hereEnable the Gemini API:
Go to API Overview → Enable Gemini.Create an API key:
Go to APIs & Services → Credentials
Click Create Credentials → API Key
Copy and store the key securely
How to Use the API Key in Kilo Code (or Cursor, etc.)
Open your IDE (VSCode, Cursor, etc).
Open Kilo Code from the Command Palette (Cmd + Shift + P).
Create a new configuration profile (e.g. gemini-vertex-paid).
Choose Google Gemini as the provider (or GCP Vertex AI).1
Paste your GCP API key.
Select model: google/gemini-2.5-pro-preview-03-25
Hit Save. Done.
Final Thought
Whether you’re using Kilo Code or hopping across tools like Windsurf, we want you to build fast, affordably, and smart. Take the $300 when you can. Use our $20 too. Just don’t say we never gave you anything.
Pro Tip: API Provider Selection
Start with Google Gemini as an API provider, though you may need to switch it to GCP Vertex AI.