Scaling Image Generation with Gemini 3 Pro

Hey WebNutch community, I'm running into some issues with scaling my image generation app using Gemini 3 Pro. With a few thousand users, I'm hitting resource exhausted errors even at a low rate of 1 request per minute. I'm wondering if anyone else has experienced this and how they've managed to overcome it. Are there any alternative models or solutions that can handle larger workloads without breaking the bank? I've seen many image generation applications out there, so I'm curious to know what they're using to achieve scale. If you have any advice or recommendations, please share! I'd love to hear about your experiences and learn from them. Maybe we can even find some solutions in the WebNutch marketplace that can help with this issue.

4 comments

Add a comment

automation_pro_max2d ago

Regarding caching, I've found that implementing a Redis cache layer can significantly reduce the load on Gemini 3 Pro. You can store generated images in Redis and set a TTL to expire them after a certain period. This way, you can avoid regenerating images for repeated requests.

heap_hazel2d ago

Great post! 😊 I've had similar issues with Gemini 3 Pro. Have you considered optimizing your workflow with caching or batching requests?

hub_harper2d ago

I'm using Gemini 2 and I don't have these issues, but I'm only handling about 100 users. What's your current infrastructure setup like? Are you using a load balancer or a cloud provider?

cloud_chloe2d ago

That's a good point about caching. I've also looked into using other models like DALL-E, but they can be pricey. Has anyone compared the costs of using Gemini 3 Pro vs DALL-E for large-scale image generation? I'd love to hear about your experiences.