GPT 5.4 API with APIDot

GPT 5.4 API examples with APIDot: cURL, Node.js, request fields, responses, and production notes.

Try on APIDot | Get API Key | API Docs | Pricing | Main Examples

Why this repo exists

OpenAI GPT 5.4 API for efficient professional task execution, spreadsheet and document analysis, tool-oriented planning, code/test handoffs, streaming, and OpenAI-compatible chat completions.

This repository turns that APIDot workflow into runnable server-side examples: a cURL request, a native Node.js example, request fields, response handling notes, and production integration guardrails.

Overview

GPT 5.4 is OpenAI's frontier model for efficient professional work, combining broad reasoning with coding and agentic strengths from the GPT-5.3-Codex line. It is designed for work that should produce an action plan, review, handoff, or structured operational answer rather than casual chat.

Choose GPT 5.4 for spreadsheet and document reasoning, legal or financial material review, coding plans, browser-operation plans, tool selection, and professional workflows that need reliable judgment without always escalating to GPT 5.5. It is the middle ground between cheaper bulk models and the most expensive long-horizon option.

On APIDot, call gpt-5.4 through OpenAI-compatible text chat. The page centers on POST /v1/chat/completions with Bearer authentication, also documents the configured Responses-compatible path, and keeps hosted Computer Use, web search, visual input UI, and file upload outside the current APIDot page contract.

Endpoint

POST /v1/chat/completions

Send OpenAI-compatible chat completions requests to GPT 5.4 for professional task execution, spreadsheet and document analysis, tool planning, and coding handoffs.

Best practices

• Use gpt-5.4 for professional tasks such as spreadsheet review, legal or finance material analysis, tool-selection plans, browser-operation handoffs, and implementation planning.
• Give the model the business objective, evidence, tool constraints, expected output format, and verification criteria instead of asking for a generic answer.
• Use /v1/chat/completions for OpenAI-compatible chat clients and APIDot playground parity; use /v1/responses only when your client already models work as Responses-style execution.
• Read assistant text from choices[0].message.content, or from data.choices[0].message.content if your response payload is wrapped.
• Use usage.prompt_tokens and usage.completion_tokens to route GPT 5.4 as a professional-work model between cheaper bulk models and GPT 5.5.
• Set stream: true only when your client is prepared to consume streaming chat completion events.

Quick start

cp .env.example .env
# Edit .env and set APIDOT_API_KEY
cd node
npm start

Use curl/generate.md when you want a copy-paste cURL request instead of Node.js.

Minimal API request

Submit to APIDot:

curl --fail-with-body --request POST \
  --url https://api.apidot.ai/v1/chat/completions \
  --header "Authorization: Bearer $APIDOT_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
  "model": "gpt-5.4",
  "messages": [
    {
      "role": "system",
      "content": "You are a professional workflow assistant."
    },
    {
      "role": "user",
      "content": "Review this quarterly spreadsheet narrative. Identify formula assumptions to verify, finance risks, and the next checks an analyst should run."
    }
  ],
  "max_tokens": 1024
}'

Primary payload:

{
  "model": "gpt-5.4",
  "messages": [
    {
      "role": "system",
      "content": "You are a professional workflow assistant."
    },
    {
      "role": "user",
      "content": "Review this quarterly spreadsheet narrative. Identify formula assumptions to verify, finance risks, and the next checks an analyst should run."
    }
  ],
  "max_tokens": 1024
}

Request variants

gpt-5.4

{
  "model": "gpt-5.4",
  "messages": [
    {
      "role": "system",
      "content": "You are a professional workflow assistant."
    },
    {
      "role": "user",
      "content": "Review this quarterly spreadsheet narrative. Identify formula assumptions to verify, finance risks, and the next checks an analyst should run."
    }
  ],
  "max_tokens": 1024
}

Request parameters

Field	Type	Required	Description
model	string	yes	Target model ID. Use gpt-5.4.
messages	object[]	yes	Conversation array in OpenAI-compatible chat format. Include the objective, evidence, tool constraints, and desired professional output.
messages[].role	string	yes	Message role. Use system, user, or assistant.
messages[].content	string	yes	Text content for the message.
temperature	number	no	Sampling temperature, usually between 0 and 2.
top_p	number	no	Nucleus sampling value, usually between 0 and 1.
max_tokens	integer	no	Maximum output tokens to return. Increase it for spreadsheet audits, professional memos, or implementation plans.
stream	boolean	no	Set true to receive streaming chat completion events.

Response fields

Field	Type	Description
id	string	Chat completion response ID.
choices[].message.content	string	Generated assistant text.
choices[].finish_reason	string	Reason generation stopped.
usage.prompt_tokens	integer	Input token count billed at the GPT 5.4 input rate.
usage.completion_tokens	integer	Output token count billed at the GPT 5.4 output rate.
usage.total_tokens	integer	Total token count across input and output.

Example response

{
  "id": "chatcmpl_1234567890",
  "object": "chat.completion",
  "model": "gpt-5.4",
  "choices": [{"index": 0, "message": {"role": "assistant", "content": "Start by checking the model assumptions, then verify the variance drivers..."}, "finish_reason": "stop"}],
  "usage": {"prompt_tokens": 126, "completion_tokens": 228, "total_tokens": 354}
}

Error classes

Code	Name	Description
401	Unauthorized	Missing, invalid, or disabled APIDot API key.
400	Invalid request	The request body is missing required fields, or the requested API path is not supported for this model.
402	Insufficient credits	The account does not have enough credits for the requested token usage.
500	Service error	The GPT 5.4 request failed or timed out.

Production notes

• Keep APIDot API keys in server-side environment variables.
• Persist selected model, request payload, user ID, and response metadata together for support and cost review.
• Avoid logging API keys, private prompts, customer data, private media URLs, or internal callback URLs.
• Validate request fields before sending traffic to APIDot.
• Retry transient network failures with backoff, but do not retry unchanged invalid payloads.

Common mistakes

• Committing a real API key or .env file.
• Calling APIDot directly from browser code instead of a server-side environment.
• Logging private prompts, customer data, API keys, or internal callback URLs.
• Reusing request fields from another model without checking this model's docs.

FAQ

Which GPT 5.4 model ID can I send?

Send gpt-5.4 in the model field.

What authentication header should I use?

Use Authorization: Bearer $APIDOT_API_KEY when calling /v1/chat/completions.

When should I use /v1/chat/completions instead of /v1/responses?

Use /v1/chat/completions for OpenAI-compatible chat clients and APIDot playground parity. Use /v1/responses when your integration already depends on Responses-style workflow state.

Does this docs page expose Computer Use, web search, or file upload UI?

No. This APIDot docs page documents text chat completions and the configured Responses path. It does not expose hosted Computer Use, web search, file upload UI, or a platform-level agent environment.

Can I stream responses?

Yes. Set stream: true to receive streaming chat completion events.

How is usage reported?

Read usage.prompt_tokens, usage.completion_tokens, and usage.total_tokens from the response for cost tracking and model routing.

Related links

• Website: https://apidot.ai
• Docs: https://apidot.ai/docs
• GPT 5.4 docs: https://apidot.ai/docs/gpt-5-4
• GPT 5.4 model page: https://apidot.ai/models/gpt-5-4
• GitHub organization: https://github.com/APIDotAI
• Main examples: https://github.com/APIDotAI/apidot-examples