Unsure about deploying a Keras model (480 MB) on Render's free tier

I’m considering deploying a Flask API with a Keras machine learning model (480 MB) on Render’s free tier for 10 days with low to very low traffic. I’m unsure about exceeding free tier limits and incurring unexpected charges. Additionally, I’d like to understand the request limit for deployed apps.

While the free tier seems appealing, I have some concerns:

  1. Given the limited free instance hours and bandwidth, is it possible to unknowingly exceed these limits and incur unexpected charges, especially with the model running for 10 days?

  2. Could you please clarify the specific limitations of the free tier, particularly regarding:

  • Is size of the model (480 MB) enough for render’s free plan?
  • Is there a limit on maximum number of requests I can send to my deployed app?

if you are on the free tier you do not need to add a payment method. the only thing that can happen once you’ve exceeded the 750 instance hours or 100 gig transfer, i think, they suspend it, but since there is no payment method associated with a free tier account there is no way to incur charges


  1. It depends on how you’re set up but in general, with a free tier and no cards added you won’t be even able to incur any charges in the first place.
  2. I would advise against using the free tier, or even the ‘Starter’ instance type, for this model size. At the very least, you should opt for the ‘Standard’ instance. Considering that all, or a significant portion, of your 480MB model will need to reside in memory, the two smallest instance types do not provide sufficient memory to support both the model and all associated processes.

Render Support, UTC+3

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.