Here, we will introduce the API usage norms and related information for our OpenAI platform.

Please note that you need to log in to the OpenAI platform using a company email to enter the company group.

Token Consumption Calculation

Generally speaking, every 2.5 to 3 English characters count as one Token (https://github.com/openai/tiktoken (opens in a new tab)), so every 1k Tokens can generate about 2800 characters. However, since the API charges for both inputs and outputs, the characters input will also be counted towards the Token consumption. Therefore, theoretically, the maximum output for every 1k Tokens is about 2000 characters, and the actual knowledge content (based on general usage of GPT software) is up to 1000 characters, or about 200 words. In other words, for the GPT-4.0 128k version, each 1 USD can generate 5000 words.