From Morning, This has been the case. Kindly look into it.
Model - Gemini 2.5 Pro
Yes, it seems this model’s failure rate is almost 50%, and failed responses also count against the quota. BTW, I’m curious why you’re still using this model. I thought no one had been using it anymore.
I like the Writing Style of it. It is Working fine till yesterday, But I think AI Studio Pro Subscribers are facing problem in all the models.
@Logan_Kilpatrick @Mustan_lokhand @Mahesh_Sutar @Jon_Matthews Kindly look into this issue. The Model is going to get deprecated by October, I wish to use it to the fullest before it’s demise.
I purchased AI Pro Subscription just to use Gemini 2.5 Pro in AI Studio. But you guys are wasting it’s potential. After 2.5 Pro whatever model doesn’t match the writing style of 2.5 pro.
Can you please resolve this issue ASAP. It is affecting my work. Kindly do the needful ASAP.
I am relying on 2.5 Pro for all my professional writing needs. This is the best model
Same here — exactly the same use case. But relax, better models will definitely come along in the future. This is actually a post-training issue. Right now, everyone is focusing post-training on coding, so that’s why things are like this. But the market for creative writing is actually huge too, and it’ll eventually get more attention.
Even if Google completely fumbles it, some other company will come along and build a better writing model. Of course, OpenAI has also kind of fumbled it, because the language style of their data annotators is… pretty questionable, to put it mildly.
Hi all, we’re aware of some serving issues with the 2.5 series models and are looking into it.
Thank you, I’ve been noticing these problems too. Just reply on when it’ll be fixed.
@Logan_Kilpatrick @Mustan_lokhand @Mahesh_Sutar @Jon_Matthews Kindly look into this issue. The Model is going to get deprecated by October, I wish to use it to the fullest before it’s demise.
We’ve applied a fix for this… 503 error rates should be far lower now.