GPT-5.3 Codex API

Billed from your GPUniq balance as you use it — no subscription, no monthly minimum, no per-seat fee. Prices refresh from the live catalog hourly.

API identifier

gpt-5.3-codex

Pass this exact string as the "model" field of your request.

Type

Chat & text

Served by OpenAI.

Context window

400,000 tokens

About roughly 300,000 words — about 600 printed pages. Prompt, conversation history and attachments all count against it.

Max output

128,000 tokens

Ceiling for a single reply. Set max_tokens below it to cap cost per request.

Accepts

Text, Images, Files (PDF and similar documents)

What you can put in the request body besides plain text.

Returns

Text

Measured speed

≈2290 tokens/sec

Rolling average of real GPUniq traffic, measured on streamed responses.

Tokenizer

GPT

Determines how your text splits into billable tokens.

Available since

2026-02-24

Upstream moderation

Yes

The vendor applies its own safety filter, which can reject a request before it reaches the model.

• Function calling: supported

  Send tool definitions, get back the call the model wants made.

• Structured outputs: supported

  Replies constrained to your JSON Schema.

• Vision input: supported

  Accepts images in the message content.

• Extended reasoning: supported

  Thinks before answering; thinking tokens bill as output.

curl https://api.gpuniq.com/v1/openai/chat/completions \
  -H "Authorization: Bearer YOUR_GPUNIQ_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.3-codex",
    "messages": [{"role": "user", "content": "Explain rate limiting in one paragraph."}]
  }'

Create a key on /chat and send it as a bearer token. The same key works across every model in the catalog, so switching models means changing one string.

include_reasoning

Returns the reasoning trace alongside the answer instead of hiding it.

max_completion_tokens

The newer name for max_tokens; both are accepted where the model lists them.

max_tokens

Hard cap on the reply length. Also your cost ceiling per request, since output is the expensive half of the bill.

reasoning

Extended thinking. The model works through the problem before answering; the thinking tokens are billed as output.

reasoning_effort

Dial for how long the model may think (typically low / medium / high). Higher settings cost more because thinking tokens are output tokens.

response_format

Selects the response shape — plain text or JSON. The weaker cousin of structured outputs: it enforces valid JSON, not your particular schema.

seed

Asks for repeatable sampling. Best effort on every vendor — it makes runs similar, it does not make them identical.

structured_outputs

Schema-constrained decoding: the reply is guaranteed to parse against the JSON Schema you supply, so no retry loop around JSON.parse.

tool_choice

Forces or forbids a tool call for one request — useful when you want a guaranteed structured answer instead of prose.

tools

Function calling. You describe callable functions in JSON Schema and the model replies with the call it wants made, which is what agent frameworks are built on.