How to Use Gemini 2.5 Pro in Cursor for Free (2026)

Cursor ships with Claude and GPT models, but those eat through your monthly request quota fast. What many developers do not realize is that you can connect Gemini 2.5 Pro to Cursor using a free Google AI Studio API key, giving you one of the best coding models available with essentially unlimited free usage.

This guide walks you through the complete setup process, from generating your free API key to configuring Cursor and optimizing your workflow.

Why Use Gemini 2.5 Pro in Cursor?

Before we set things up, here is why Gemini 2.5 Pro is worth adding to your Cursor configuration:

Feature	Gemini 2.5 Pro	Claude Sonnet (Cursor default)	GPT-4o (Cursor default)
Context window	1,000,000 tokens	200,000 tokens	128,000 tokens
Free API tier	Yes (5 RPM)	No	No
Coding benchmarks	Top tier	Top tier	Strong
Multimodal input	Text, image, video	Text, image	Text, image
Cost via Cursor	$0 (own key)	Uses Cursor quota	Uses Cursor quota

The key advantage: Gemini 2.5 Pro requests made with your own API key do not count against your Cursor Pro or free tier limits. You keep your 500 (Pro) or 50 (free) premium requests intact for Claude and GPT, while getting unlimited Gemini 2.5 Pro usage within Google's free rate limits.

Step 1: Get a Free Google AI Studio API Key

Go to aistudio.google.com/apikey
Sign in with your Google account
Click Create API Key
Select an existing Google Cloud project or create a new one
Copy the generated key and save it somewhere secure

The API key is completely free. No credit card is required, and there is no trial period -- it remains free indefinitely within the rate limits.

Free Tier Rate Limits

Limit	Amount
Requests per minute	5
Tokens per minute	250,000
Requests per day	100
Context window	1,000,000 tokens

For most individual developer workflows in Cursor, 5 requests per minute and 100 requests per day is more than enough. If you need higher limits, you can add billing to your Google Cloud project and pay per token.

Step 2: Configure Gemini 2.5 Pro in Cursor

Cursor supports custom models through its OpenAI-compatible API configuration. Google provides an OpenAI-compatible endpoint for Gemini models, making integration straightforward.

Method A: Via Cursor Settings UI

Open Cursor
Go to Settings (gear icon) > Models
Scroll to OpenAI API Key or Custom Models
Click Add Model or Add Custom Endpoint
Enter the following:
- Model Name: gemini-2.5-pro
- Base URL: https://generativelanguage.googleapis.com/v1beta/openai
- API Key: Your Google AI Studio API key
Click Save and verify the model appears in your model selector

Method B: Via Cursor Configuration File

You can also configure it directly in the Cursor settings JSON. Open the command palette (Cmd+Shift+P / Ctrl+Shift+P) and search for "Open Settings JSON":

{
  "cursor.aiModels": {
    "custom": [
      {
        "name": "gemini-2.5-pro",
        "baseUrl": "https://generativelanguage.googleapis.com/v1beta/openai",
        "apiKey": "YOUR_GOOGLE_AI_STUDIO_KEY",
        "model": "gemini-2.5-pro",
        "contextLength": 1000000
      }
    ]
  }
}

Step 3: Verify the Connection

After configuration, test that Gemini 2.5 Pro is working:

Open a new Cursor chat panel (Cmd+L / Ctrl+L)
In the model dropdown at the top, select gemini-2.5-pro
Type a test prompt: "Write a Python function to calculate Fibonacci numbers"
Confirm you get a response

If you see an error, double-check:

Your API key is correct and active
The base URL ends with /openai (not /v1beta alone)
You have not exceeded the free rate limits

Step 4: Use Gemini 2.5 Pro in Different Cursor Modes

Chat Mode

Gemini 2.5 Pro works seamlessly in Cursor's chat panel. Select it from the model dropdown and use it for:

Explaining code
Debugging errors
Generating boilerplate
Answering architecture questions

Composer / Agent Mode

In Agent mode, Gemini 2.5 Pro can make multi-file edits. Its 1M token context window is especially useful here because it can hold your entire codebase in context:

@codebase Refactor the authentication module to use JWT tokens
instead of session-based auth. Update all related files.

With 1M tokens of context, Gemini 2.5 Pro can reason over significantly more code than Claude or GPT-4o in a single turn.

Inline Editing

For inline edits (Cmd+K / Ctrl+K), Gemini 2.5 Pro works as a drop-in replacement. Highlight code, press the shortcut, and type your edit instruction. The model will suggest changes inline.

Step 5: Optimize Your Workflow

Use Gemini for Large Context, Claude for Precision

A practical strategy is to use both models:

Task	Best Model	Why
Multi-file refactors	Gemini 2.5 Pro	1M context fits entire repos
Complex debugging	Claude Sonnet	Strong reasoning
Boilerplate generation	Gemini 2.5 Pro	Free, fast
Architecture decisions	Either	Both are strong
Quick one-liners	Gemini 2.5 Flash	Fastest, free

Switch between models using the dropdown in Cursor's chat panel. Using Gemini for routine tasks preserves your Cursor premium requests for when you specifically need Claude or GPT.

Add Gemini 2.5 Flash Too

While you are adding custom models, also add Gemini 2.5 Flash for fast, lightweight tasks:

{
  "name": "gemini-2.5-flash",
  "baseUrl": "https://generativelanguage.googleapis.com/v1beta/openai",
  "apiKey": "YOUR_GOOGLE_AI_STUDIO_KEY",
  "model": "gemini-2.5-flash",
  "contextLength": 1000000
}

Gemini 2.5 Flash has a more generous free tier (15 RPM, 500 RPD) and responds faster. Use it for autocomplete-style tasks and save Gemini 2.5 Pro for complex reasoning.

System Prompts

Cursor lets you set custom system prompts per model in .cursorrules or the settings. A good system prompt for Gemini 2.5 Pro in a coding context:

You are a senior software engineer. Write clean, production-ready code.
Follow existing code conventions in the project. Prefer simple solutions.
Always explain your reasoning briefly before writing code.
If the request is ambiguous, ask for clarification instead of guessing.

Troubleshooting Common Issues

"Model not found" error Make sure the model name is exactly gemini-2.5-pro (not gemini-2.5-pro-latest or other variants). Check the Google AI Studio docs for current model IDs.

Slow responses The free tier has a 5 RPM limit. If you are hitting it, wait a moment between requests or switch to Gemini 2.5 Flash for faster responses.

Responses cut off mid-sentence Increase the max output tokens in Cursor's model settings. Gemini 2.5 Pro supports up to 65,536 output tokens.

API key not working Ensure you generated the key at aistudio.google.com/apikey, not the Google Cloud Console. AI Studio keys work directly; Cloud Console keys require additional OAuth setup.

Cursor does not show the model in dropdown Restart Cursor after adding a custom model. If it still does not appear, check that your configuration JSON is valid (no trailing commas, correct structure).

Frequently Asked Questions

Does using my own API key void Cursor's terms of service? No. Cursor officially supports custom API keys and custom model endpoints. This is a supported feature.

Is the Google AI Studio API key really free forever? Google provides a free tier with rate limits and has maintained it since Gemini's launch. There is no trial period and no credit card required. Google may adjust limits over time, but free access has been consistent.

Can I use this on Cursor's free Hobby plan? Yes. This setup works on any Cursor plan, including the free tier. Your Gemini requests are completely independent of your Cursor subscription.

How does latency compare to Cursor's built-in models? Gemini 2.5 Pro responses typically take 2-5 seconds for medium-length outputs. This is comparable to Claude Sonnet through Cursor's servers. Gemini 2.5 Flash is noticeably faster at 1-2 seconds.

Wrapping Up

Adding Gemini 2.5 Pro to Cursor gives you a free, high-quality coding model with the largest context window available. The setup takes under five minutes, and you immediately get unlimited access within the free rate limits -- no impact on your Cursor subscription quota.

For developers building applications with AI-generated media, Hypereal AI offers a unified API for image generation, video generation, and talking avatars. Sign up free with no credit card to get started.