Bug Report: Gemini 2.5 Pro Degradation

Bug Reports
1

Bug Report: Gemini 2.5 Pro Degradation

Date: April 24, 2025

Model: Gemini 2.5 Pro

Description:

I am reporting a significant and concerning degradation in the performance and capabilities of Gemini 2.5 Pro. The model, which was previously performing adequately, has recently exhibited a severe decline in its ability to understand and execute basic instructions, particularly those related to system-level operations.

Specific Issues:

  1. Inability to Execute Terminal Commands: Gemini 2.5 Pro frequently fails to recognize and execute terminal commands, even when they are clearly and explicitly requested. It often misunderstands the intent, providing natural language responses instead of the expected command-line execution.
  2. Misinterpretation of Folders: The model incorrectly interprets requests to create folders, often attempting to create files with the folder names instead. This indicates a fundamental misunderstanding of file system structure and operations.
  3. Complete Refusal to Respond: In some instances, Gemini 2.5 Pro becomes unresponsive and refuses to perform any tasks, regardless of the complexity or simplicity of the request. This complete lack of response makes the model unusable.
  4. General Stupidity and Downgraded Performance: Beyond the specific issues listed above, the model exhibits a general decline in overall intelligence. It struggles with tasks it previously handled with ease, providing incorrect, nonsensical, or irrelevant responses. The model’s behaviour can only be described as “stupid” at times.
  5. Gemini 2.5 Flash Issue: When using Gemini 2.5 Flash once in a chat, subsequent attempts to use it result in the error message: ““We’re having trouble connecting to the model provider. This might be temporary - please try again in a moment.””

Impact:

These issues collectively render Gemini 2.5 Pro significantly less useful, if not entirely unusable, for any task that requires a basic understanding of system-level operations or accurate instruction following. The degradation in performance is substantial and severely impacts productivity. The fact that this issue “isn’t being resolved” is unacceptable.

1 Like
2

Hi, please open your console (in the command palette, Developer: Toggle Developer Tools) and send us screenshots of any errors you see here!

And you can provide information about your device by clicking Help—About to make it easier for officials to find the problem.

Please share some request IDs so we can check on these queries. If you’re in privacy mode, would be great if you can temporarily disable, perform a query, and send us the request id

3

I’ve noticed it too. Over the past two days, it’s been frequently writing regular expressions incorrectly, and I’ve had to use version 4.1 to fix its mistakes. Isn’t this supposed to be the top-ranked model? I honestly can’t tell—it feels like it’s only at GPT-3.5 level based on my recent experience.

2 Likes
4

Yes, still an ongoing issue.

They have closed this thread: Gemini 2.5 Issues - Looping, Bad Performance, Wrong Language, etc - #18 by ogmios2

But it is still completely useless today Apr 25. It is still behaving exactly as before in the closed thread.

A complete mess, waste of time and money.

1 Like
5

It’s a waste of money when they don’t fix stuff like this, there was no issue before but now there is and they have said it’s fixed but clearly isn’t and that I’m not the only one having this issue.

1 Like