Google Gemini 2.5 Pro is Back on the Free API Tier with 100 Daily Requests

Other Questions API
1

Google has reintroduced the Gemini 2.5 Pro model in their free API tier, now allowing 100 requests per day. This is a helpful change considering that Copilot currently limits the use of premium models to just 300 requests a month.

Available features:

  • 100 requests per day for Gemini 2.5 Pro
  • 250 requests per day for Gemini 2.5 Flash (a decrease from the previous 500)

How to set it up:

  1. Obtain your API key from Google AI Studio.
  2. Use the Insiders version of Copilot.
  3. Go to the model management section.
  4. Input your Gemini API key.
  5. Create a new model named gemini-2.5-pro.
  6. Choose the “Advanced” settings option.
  7. Set token limits as follows:
    • Input tokens: 819600
    • Output tokens: 65536
  8. Make sure to enable features for tool calling and vision.

Key reminders:

  • Ensure you select the official Gemini 2.5 Pro model and not the preview version.
  • Select it from the “Other Models” list to avoid deducting from your 300 premium requests.
  • Agent mode is functional with custom API keys now.

Bonus tip: You can also add the gemini-2.5-flash model for an additional 250 daily requests and the gemini-2.5-flash-lite-preview-06-17 model for 1000 requests daily. This setup results in approximately 3000 requests a month in addition to your 300 premium requests, offering a more robust performance compared to the free 4.1 model.

2

Been testing this for a few days - the performance boost over free tier models is definitely noticeable. The 100 daily limit works for most tasks, but I’m bummed they cut Flash from 500 to 250 requests. API responses are way faster than using Gemini’s web interface. Having all three models is pretty flexible - I throw complex tasks at Pro, general queries at Flash, and simple stuff at Flash-lite. Setup was easy following those steps; just ensure your API key has the right permissions.

3

Finally got around to implementing this after procrastinating for weeks. The token limits in the setup are crucial - I used default values at first and kept hitting walls with longer conversations. What surprised me most is how much better the vision capabilities work through the API versus the web interface. Processing images feels way more reliable now. Heads up - the daily quota resets at midnight UTC, not your local timezone, which caught me off guard. For anyone hesitant about switching from the standard models, the quality difference is worth the setup time.

4

omg this is huge! :heart_eyes: i kept runnning into that 300 limit, it’s sooo annoying. got mine set up now but had to refresh a bunch of times to see gemini 2.5 pro in the “other models.” flash-lite is actually pretty good for most stuff, who knew?