Restore Gemini 3.1 Pro or Provide Access to Argon for Jio Subscribers and Students

Google Antigravity
1

Hi Antigravity Team,

Following the recent update, I noticed that Gemini 3.1 Pro has been quietly removed from the model selector for Google AI Pro subscribers, leaving only Gemini 3.8 Flash available under the Gemini lineup.

Although Flash 3.8 may perform well on synthetic benchmarks, my experience suggests that it is not yet reliable enough for real-world engineering workflows. While working on an active, medium-sized project, I encountered several serious issues:

  • Hallucinations and Unintended Code Changes: The model frequently hallucinated APIs and modified unrelated files, introducing regressions across the codebase. Instead of resolving issues, it broke existing functionality and left the project in a chaotic state.

  • Infinite Loops and Quota Consumption: It repeatedly got stuck in read-and-think loops without making meaningful changes, while consuming my quota. Eventually, I had to revert all its changes using Git and switch back to Gemini 3.1 Pro to restore the project.

Gemini 3.1 Pro offered a much better balance of architectural reasoning, context stability, reliable execution, and cost. Removing it limits the usefulness of Antigravity for serious development work.

I request that you restore Gemini 3.1 Pro or provide access to Argon as a reasoning model, especially for Jio subscribers and students. At the very least, please consider allowing users to route complex reasoning, debugging, and refactoring tasks through more capable Pro-tier models.

Many Jio subscribers and students have subscriptions with several months remaining. Switching to another paid subscription may require cancelling an existing Jio plan, potentially causing the remaining subscription period to go to waste. We should not have to choose between losing the value of an existing subscription and accessing the models we need for reliable development.

Please consider providing continued access to Gemini 3.1 Pro or Argon for Jio subscribers and student plans. jio users and students deserve dependable reasoning models for real-world coding tasks, not just models that perform well on benchmarks.

Thank you for your time and consideration. I hope you will take this feedback seriously and provide a solution that supports the needs of developers, students, and existing subscribers.

Best regards,
A Concerned Antigravity User