[noobcoder]

Is it worth it to host this on an EC2 which might take ~1.5$ per hour (on demand) than running GPT3.5 API for this purpose? What is the breakeven number of queries (~2000 tokens/query) to justify the hosting of such model?