Is there currently an issue with the Gemini API for image generation? Latency suddenly increased from 9–40s to up to 5 minutes

Hi everyone,

I’m trying to understand whether there is currently an issue with the Gemini API for image generation, or whether this is only happening on my side.

I’m building an application that uses Gemini image generation, and I log the generation time for every request in my backend.

Under normal conditions, image generation usually takes between 9 and 40 seconds in my app, depending on the request and whether I generate in 1K or 2K.

Since around midday today, latency has changed dramatically:

  • some requests are still fast

  • some requests are now taking much longer than usual

  • in some cases, generation can take up to 5 minutes

This is a sudden change compared to the behavior I had earlier today.

At first I thought it might be caused by my own backend timeout settings, but I also tested in Google AI Studio / Playground and observed very long generation times there as well, which makes me suspect a broader issue with the API or the image generation service.

My setup:

  • Gemini API

  • image generation in 1K and 2K

  • backend logs generation duration for every request

  • normal observed latency before today: 9 to 40 seconds

  • current observed latency since midday: sometimes up to 5 minutes

Is anyone else seeing the same thing today?

Is there currently a known issue or degradation on the Gemini API for image generation?

Thanks

1 Like

Yesy exactily the same constat…. more than 150sec for 1k

(post deleted by author)

Same issues all morning and currently as well. I tried changing some api call settings but nothing helps. I’m Tier 3 so this is very unusual for me. Hope they fix it soon.

It is truly vexing to remain ignorant of the service’s rationale and its current state of operation.