Gemini 2.5 Pro - Reduced performance, frequent early response terminations, request errors

Feedback
1

Hi,

I’ve been noticing a lot of issues with 2.5. Pro lately. It is one of my favourite models, but recently has been having a lot of issues with early response terminations, request errors (“We’re having trouble connecting to the model provider. This might be temporary - please try again in a moment.”), and so forth.

This is quite annoying as it’s unpredictable, and continuing a response often leads to failure anyway. It also rarely ever calls MCP tools, despite clear instructions, where in GPT-5 almost always follows instructions perfectly, but at a higher pricing.

I’d like to report this in since I’d love to continue using Pro 2.5. but atm, I essentially have to start all my responses with GPT-5 and then continue using 2.5. Pro as otherwise Gemini is too unreliable. Thanks! Hope this reaches you guys and you can figure out what’s up, whether it’s on Google or something.

2

Hey, thanks for the report. Yes, this issue is on Google’s end, and the team is already working on it with them.

1 Like
3

I see, thank you so much! Hoping this gets fixed soon :slight_smile: