This message was deleted.

bentoctl does support deploying to aws lambda although I haven’t experiment with stable diffusion model on lambda. BentoCloud does offer serverless deployment with gpu support, which works great for this type of models, although it is still under private beta.

@Chaoyu what do you think about decreasing cost? I just wanna to get the approach when I have to pay If I use it, Is it possible?

Yes that’s one of the key design goals for BentoCloud’s serverless architecture - when a model is not actively being used, it scales down to zero without occupying the CPU/GPU resources, which helps with cost saving for our customers