update readme

README.md

@@ -93,6 +93,16 @@ request.
   [here](https://platform.openai.com/account/api-keys). Please add this key to
   your GitHub Action secrets.
 
+### Models: `gpt-4` and `gpt-3.5-turbo`
+
+At FluxNinja, we use `gpt-3.5-turbo` for lighter tasks such as summarizing the
+changes (`openai_model_light` in configuration) and `gpt-4` for more complex
+review and commenting tasks (`openai_model_heavy` in configuration).
+
+Costs: `gpt-3.5-turbo` is dirt cheap. `gpt-4` is orders of magnitude more
+expensive, but the results are vastly superior. We are typically spending $50 a
+day for a 20 developer team with `gpt-4` based review and commenting.
+
 ### Prompts & Configuration
 
 See: [action.yml](./action.yml)

action.yml

@@ -102,10 +102,10 @@ inputs:
         - Complexity
         - Optimization
 
-      In general, refrain from making noisy comments such as on minor 
-      code style issues, compliments, missing documentation, etc. 
-      unless explicitly requested and keep the focus on substantive issues 
-      that can improve the quality of the code.
+      Refrain from addressing minor code style issues, compliments, or missing 
+      comments/documentation, unless explicitly requested. Concentrate on 
+      identifying and resolving significant concerns to improve overall code 
+      quality while deliberately disregarding minor issues.
   summarize_file_diff:
     required: false
     description: 'The prompt for file'

src/review.ts

@@ -415,13 +415,12 @@ Important instructions:
   chain when reviewing the new hunk.
 - Use Markdown format for review comment text.
 - Fenced code blocks must be used for new content and replacement 
-  code/text snippets. 
-- If you have a suggestion for replacing the code between the specified 
-  line number range, please include it in a fenced code block with the 
-  \`suggestion\` as the language identifier. The line number range 
+  code/text snippets.
+- If needed, provide a replacement suggestion using fenced code blocks 
+  with the \`suggestion\` as the language identifier. The line number range 
   in the review section must map exactly to the line number range (inclusive) 
-  that need to be replaced within a new_hunk_for_review. For instance, if 2 
-  lines of code in a hunk need to be replaced with 15 lines 
+  that need to be replaced within a new_hunk_for_review.
+  For instance, if 2 lines of code in a hunk need to be replaced with 15 lines 
   of code, the line number range must be those exact 2 lines. If an entire hunk 
   need to be replaced with new code, then the line number range must be the 
   entire hunk. Replacement suggestions should be complete units that can be