Chat Completions API

Creator
Creator
Seonglae ChoSeonglae Cho
Created
Created
2023 Jul 9 10:48
Editor
Edited
Edited
2025 Jan 16 13:26
Refs
Refs
The completions API endpoint received its final update in July 2023 and has a different interface than the new chat completions endpoint. Instead of the input being a list of messages, the input is a freeform text string called a prompt.
JSON mode will not guarantee the output matches any specific schema, only that it is valid and parses without errors.
accounts for 97% of API GPT usage
Chat Completions API Notion
 
 
 
 

Log probability as argument

OpenAI Platform
Explore developer resources, tutorials, API docs, and dynamic examples to get the most out of OpenAI's platform.
OpenAI Platform

Predicted output

Any tokens provided in predictions that are not part of the final completion will be charged at completion token rates. This means the more an output differs from the prediction, the more it will cost. Using Predicted Outputs can reduce processing time.
OpenAI Platform
Explore developer resources, tutorials, API docs, and dynamic examples to get the most out of OpenAI's platform.
OpenAI Platform

Models

OpenAI Platform
Explore developer resources, tutorials, API docs, and dynamic examples to get the most out of OpenAI's platform.
OpenAI Platform
Prompt Caching for cost reduction and Latency impprovement (Automatic Application)
Prompt Caching in the API
Offering automatic discounts on inputs that the model has recently seen
Prompt Caching in the API
OpenAI Prompt Caching in GPT 4o and o1: How Does It Compare To Claude Prompt Caching? - Bind AI
OpenAI recently introduced prompt caching features as a part of its annual DevDay announcements. Prompt caching—which OpenAI claims can benefit users with a 50% discount on inputs—will now applied to various models, including GPT-4o and its mini versions. Unsurprisingly, this has generated excitement among developers, with many already drawing comparisons between OpenAI's and Claude's prompt
OpenAI Prompt Caching in GPT 4o and o1: How Does It Compare To Claude Prompt Caching? - Bind AI
 

Recommendations