Hello all,

I have a trained Tensorflow model ready to be deployed and start doing some Object Detection predictions. This is for a mobile app I am building that will eventually make API calls to this particular service. Running Tensorflow on each mobile device is not an option so I want to deploy it somewhere.

I was planning to deploy it on Google Cloud but the costs are a little high and I’m just bootstrapping this particular business and also have to pay for other services.

I would like to know which Digital Ocean Droplet would be best for this use case.
As a comparison, and if I understood their pricing guide correctly, a Google Cloud node that would support 2.64 prediction queries per second would cost around $1300 per month. It seems these nodes use NVIDIA® P100 GPUs.

I know there’re no GPUs in Digital Ocean (correct me if I’m wrong) so I would be opting for the best tradeoff in terms of cost and performance.

As mentioned above, the model is already trained and I just need to do the predictions.

Thanks for your help!