Hi everyone,
I wanted to share some feedback on the plan to deprecate Gemini 2.5 Pro.
After openai deprecate 4o, I turned to Gemini. My workflow heavily relies on 2.5 Pro, not just for its power, but for its predictable style. I’ve spent months optimizing my prompts and processes around the specific way it responds. I know what to expect, and that predictability is key for my productivity.
Switching to 3.0 isn’t a simple upgrade for me. The response style is so different that it breaks my existing workflows, forcing me to start over on prompt engineering. This is a huge hidden cost for users who rely on the platform for serious work.
The recent cut in the 2.5 Pro usage limits in AI Studio is making this worse. It feels like a forced switch to a tool that doesn’t fit my established process yet, and it’s very disruptive.
This brings me to my main point: a model’s consistent behavior is a feature. It’s what allows us to build reliable tools and processes on top of Gemini. If the core behavior changes drastically with every new version, it makes it hard to trust the platform for long-term projects. We end up building on shifting sand.
I’m wondering if the team has considered options to address this? Maybe something like:
Long-Term Support versions of popular models for those who need stability. This is standard practice for a reason.
A way to maybe keep the “style” of an old model.
I think focusing on this behavioral stability will be a huge reason for users to stick with Google’s platform over the long run.