Skip to main content

openai/gpt-5.4-mini/chat

GPT-5.4 Mini examines short bug reports, reads screenshots alongside source excerpts, and returns the next diagnostic step as text or a function call.

Input $0.60Output $3.60per 1 million tokens20% below the publisher's list priceCached input $0.06OpenAI official priceInput $0.75Output $4.50

Workbench

GPT-5.4 Mini

What can GPT-5.4 Mini help with?

Related models and next tasks

Missing the model you need?

Triage a failing test

Key numbers

Input limit (tokens)
272k

Token budget

Input window
272,000
tokens in one request
Longest reply
128,000
47% of the window

reasoning_effort

  • nonedefault
  • low
  • medium
  • high
  • xhigh
messages
up to 256
tools
up to 128

What it does

Triage a failing test

Give the model the assertion, the failing input and the two functions involved. Ask which observation would distinguish a conversion error from stale data. It can suggest a check without access to the whole repository.

What a call to GPT-5.4 Mini looks like

Request bodies from this family's own examples: roles, parameters and attachments exactly as they are sent. Each one is checked against the model's input schema before it is published.

  • Find where precision is lost

    user

    A test expects 1.25, but parseRate("1.25") followed by saveRate returns 1. Which observation would tell us which function lost the fraction? Answer in at most 75 words.

    max_tokens384reasoning_effortnone
  • Test valid and invalid limits

    user

    An endpoint accepts integer quantities from 1 to 4. Give the smallest set of boundary inputs that tests both valid limits and the immediately invalid values. Answer in at most 75 words.

    max_tokens384reasoning_effortnone
  • Honor Retry-After

    user

    A loop retries immediately after every HTTP 429. Describe a small patch that honors Retry-After and limits attempts. Answer in at most 75 words.

    max_tokens384reasoning_effortnone

Where it fits

  • From screenshot to reproduction

    Pair the visible error with the relevant handler. Mark what the screenshot actually shows, and ask the model to keep its explanation separate from evidence about the request path.

Model reference

GPT-5.4 Mini API reference

How to call the GPT-5.4 Mini API

  1. Return a check to the caller

    Define a function that looks up one test by name. Validate the returned arguments in your application, run the lookup there, then return its result so Mini can revise the diagnosis.

GPT-5.4 Mini API questions

All GPT models18