Comment by Alan Carlos on "Building a Serverless Chat App with Contextual Note Integration Featuring Amazon Bedrock, the OpenAI API, and SST"

CommentBuilding a Serverless Chat App with Contextual Note Integration Featuring Amazon Bedrock, the OpenAI API, and SST

Alan Carlos

Jul 13, 2024

I can't understand why pay twice: one to use openai and another to bedrock. Why don't setup this code locally/common host and just call the openai API?

Alan Carlos

Jul 13, 2024

I am really trying to understand this.

Tobias Schmidt

Freelance Fullstack Software Engineer

Jul 13, 2024

Hi Alan! With the AWS free tier, I’d argue that it’s basically impossible to get charged even $0.01 if it’s only used by yourself 😊 (except for the costs by the LLMs in/output tokens).

Also, it’s mainly a learning project. It should demonstrate how easily you could start building a chat bot that’s deployed in the cloud 📚

Alan Carlos

Jul 14, 2024

I see. Currently, I've built my chatbot (Flask + OpenAI API) to respond to users' questions about the company I work for. It's already functioning in a testing phase. I'm using RAG to analyze company documents. For each question/response, I'm averaging U$ 0.0006 using gpt3.5-turbo. I expect this application to handle around 20,000 interactions per month, costing approximately $12 per month. I believe this is a very low cost. Do you think using Bedrock could be cheaper? Oh, I forgot to congratulate you on the post.

Search Hashnode