openai-pricing

================

This repository hosts an unofficial library for OpenAI pricing calculations. Its purpose is to simplify the prediction of costs when using GPT models. The data is obtained from https://gptforwork.com/tools/openai-chatgpt-api-pricing-calculator.

Library usage

The module provides just one class, OpneAIPricing, which loads the pricing data when instanced and answers about max tokens and costs per tokens (prompt and completion tokens).

Currently, it supports the following models:

• GPT-4 32k
• GPT-4
• GPT-3.5 16k
• GPT-3.5
• Davinci
• Curie
• Babbage
• Ada

This list can be obtained by calling OpenAIPricing.list_available_models().

Example

from openai_pricing import OpenAIPricing

p = OpenAIPricing()

p.list_available_models()


model = 'gpt-4-32k'
tokens = 1000 # default value for tokens is 1000 


model_name = p.model_name(model) # get model name

max_tokens = p.get_max_tokens(model) # get the max tokens for a model 

price_prompt = p.get_price(model, type='prompt', tokens=1000) # get the current price per 1,000 tokens in prompt 

price_completion = p.get_price(model, type='completion', tokens=1000) # get the current price per 1,000 tokens in completion  


print('The model', model_name)

print('accepts up to', max_tokens,'tokens.')

print('It costs $', price_prompt,'per 1,000 prompt tokens')

print('and $', price_completion, 'per 1,000 completion tokens.')

# Get an average price per requests with  "p.get_price(model, type='both')". "both" is the default value for the "type" argument and 1000 is the default value for the "tokens" argument.
print('The average value for request is', p.get_price(model, type='both'),'per 1,000 tokens (500 tokens in prompt and 500 pormpt in completion.')

Output:

+--------------+-------------------+
| Model name   | Technical name    |
|--------------+-------------------|
| GPT-4 32k    | gpt-4-32k         |
| GPT-4        | gpt-4             |
| GPT-3.5 16k  | gpt-3.5-turbo-16k |
| GPT-3.5      | gpt-3.5-turbo     |
| Davinci      | text-davinci-003  |
| Curie        | text-curie-001    |
| Babbage      | text-babbage-001  |
| Ada          | text-ada-001      |
+--------------+-------------------+
The model GPT-4 32k
accepts up to 32768 tokens.
It costs $ 0.06 per 1,000 prompt tokens
and $ 0.12 per 1,000 completion tokens.
The average value for request is 0.09 per 1,000 tokens (500 tokens in prompt and 500 pormpt in completion.