openai/gpt-5.4-mini/chatGPT-5.4 Mini examines short bug reports, reads screenshots alongside source excerpts, and returns the next diagnostic step as text or a function call.
Workbench
These examples were written for this endpoint. Replace the prompt and sample media URLs with your own accessible inputs, then check the current quote and model availability before submitting.
- Model reply
Check the value after **each step**: - `parseRate("1.25")` - then `saveRate(parsed)` If `parseRate("1.25")` already gives `1`, then **parseRate** lost the fraction. If it returns `1.25` but `saveRate` returns `1`, then **saveRate** lost it. So the key observation is: **inspect the intermediate parsed value before saving**.
Find where precision is lost - 384
- none
- Model reply
Test these boundary inputs: **0, 1, 4, 5**. - **1** and **4** are the valid lower and upper limits. - **0** is just below the valid range. - **5** is just above the valid range. This is the smallest set that covers both boundary values and their immediately invalid neighbors.
Test valid and invalid limits - 384
- none
- Model reply
Add two guards: 1. On 429, read `Retry-After` and sleep that long before retrying: ```python delay = int(resp.headers.get("Retry-After", "1")) time.sleep(delay) ``` 2. Cap retries with a counter: ```python for attempt in range(MAX_RETRIES): resp = request() if resp.status_code != 429: return resp raise RuntimeError("too many 429s") ``` This avoids hammering the server and stops infinite loops.
Honor Retry-After - 384
- none
This example starts at Input $0.60 / Output $3.60 per 1 million tokens. The Playground updates the estimate when you change the settings.
