while it is true that the value $20.50 does exist on the pricing page, one would be taking it out of context if one were to argue that the service costs that much per hour. contrarily, Bedrock is substantially a pay-per-use pricing model with different quotes for different individual service offering. But in all cases the pricing is in the pennies per thousand X. The value $20.50 comes up w regard to fine tuning the Titan model, and furthermore, using the on-demand pricing scheme. So first, you should take note that AWS actually bills on microseconds of usage, and so you'd need pretty big models (and a pretty specific use case) in order to arrive at a situation where you could -- in theory -- incur costs of $20+ dollars. Additionally, you'd be incurring these costs when and only when you're training a model; not when users are using your model. hope that helps.
Wow, superinteresante ! muy buena información 👍
Good job bro!
Thank you! :D
Can you make a video for using bedrock API in realtime usecase
great idea, thanks! i'll try to create something during the month of november.
Embeddings are supported in Bedrock as well.
thank you! :)
I heard it’s 20 bucks an hour to use. Is this true?
while it is true that the value $20.50 does exist on the pricing page, one would be taking it out of context if one were to argue that the service costs that much per hour. contrarily, Bedrock is substantially a pay-per-use pricing model with different quotes for different individual service offering. But in all cases the pricing is in the pennies per thousand X. The value $20.50 comes up w regard to fine tuning the Titan model, and furthermore, using the on-demand pricing scheme. So first, you should take note that AWS actually bills on microseconds of usage, and so you'd need pretty big models (and a pretty specific use case) in order to arrive at a situation where you could -- in theory -- incur costs of $20+ dollars. Additionally, you'd be incurring these costs when and only when you're training a model; not when users are using your model. hope that helps.
@@FullStackWithLawrence Thank you for the explanation. I appreciate it.