How can I differentiate between code blocks and text in Text GPT API responses?

Identifying Code and Text in API Responses

The Text API (or text completions endpoint) returns a single string in the response's content. Unlike ChatGPT's UI—which automatically renders markdown (e.g., highlighting code blocks enclosed in triple backticks)—the API itself does not provide extra metadata that separates code from regular text.

If you need to distinguish between code and narrative content, you typically have to:

1. Parse Markdown Formatting:
   Look for markdown indicators in the returned string. For instance, code blocks are usually formatted as:

   ```language  
   // code here  
   

   You can write a parser (or use an existing markdown parser) to extract these sections from the text.
   
   

2. Custom Formatting via Function Calling:
   If you want more structured outputs, consider leveraging the Function Calling feature. You can instruct the model to output responses in a specific JSON format that separates code and text into distinct fields.

In summary, while the API response itself does not explicitly tag code versus text, employing markdown parsing or custom output instructions can achieve that separation.