GPT-5.6 Terra · Model docs

Text

5.6 Terra technical docs

Endpoint, auth header, parameters, prices, and tested request examples for GPT-5.6 Terra.

Endpoint

Endpoint and pricing

POST /responses Endpoint https://api.imole.app/v1 Base URL Same model ID gpt-5.6-terra for every tier Fixed per request Availability shown per tier

Thinking uses reasoning.effort. Pro adds reasoning.mode: "pro". Every tier keeps the same model id and the same fixed price: Luna 5 cauris, Terra 10 cauris, or Sol 20 cauris per request. Base and Thinking are available in API and Agent; Pro is API only. Imole does not impose a token ceiling.

Base 10 cauris · API + Agent

A request without reasoning is Base/none. Low effort uses the same fixed price.

Omit reasoning, or use reasoning.effort: "none" | "low"

Thinking 10 cauris · API + Agent

Thinking selects standard reasoning with reasoning.effort at the same fixed model price.

reasoning.effort: "medium" | "high"

Thinking Max 10 cauris · API + Agent

Thinking Max selects the highest standard reasoning efforts at the same fixed model price.

reasoning.effort: "xhigh" | "max"

Pro 10 cauris · API only

Pro adds reasoning.mode set to "pro" without changing the fixed model price.

reasoning.mode: "pro" · effort: "medium" | "high"

Pro Max 10 cauris · API only

Pro Max combines Pro mode with the highest reasoning efforts at the same fixed model price.

reasoning.mode: "pro" · effort: "xhigh" | "max"

Headers

Request headers

Authorization required

Bearer Imole API key.

Content-Type required

Use application/json for JSON requests.

Parameters

Request parameters

model required

Use the page model id for Base, Thinking, and Pro. The model id never changes between tiers.

gpt-5.6-terra

input required

String input or Responses input array.

messages

Legacy chat-style array. The router converts it to Responses input.

instructions

System or developer instruction inserted ahead of the user task.

max_output_tokens

Optional native Responses output control. It is not required, and Imole does not impose a token or output ceiling.

reasoning

Omit reasoning for Base/none. Thinking uses reasoning.effort. Pro adds reasoning.mode set to "pro" while keeping the same model id.

effort: none, low, medium, high, xhigh, max · mode: pro

temperature

Sampling randomness when the model supports it.

top_p

Nucleus sampling control.

stream

Return server-sent events when supported.

true, false

tools

Tool definitions for tool-enabled Responses calls.

tool_choice

Tool selection policy.

auto, none, required

parallel_tool_calls

Allow multiple tool calls in one turn.

true, false

text.format

Structured text format controls for Responses.

response_format

Compatibility field for JSON-style output requests.

metadata

Optional request metadata.

Examples

Tested request examples

Base, Thinking, and Pro use the same model id. Only the reasoning object changes.

Base · no reasoning object

curl -X POST "https://api.imole.app/v1/responses" \
  -H "Authorization: Bearer $IMOLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5.6-terra",
  "input": "Write a concise product update announcing a faster API workflow.",
  "instructions": "Answer in three short bullet points."
}'

Thinking · reasoning.effort

curl -X POST "https://api.imole.app/v1/responses" \
  -H "Authorization: Bearer $IMOLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5.6-terra",
  "input": "Compare exponential backoff and fixed-delay retries for an API client, then recommend one.",
  "reasoning": {
    "effort": "medium"
  }
}'

Pro · reasoning.mode

curl -X POST "https://api.imole.app/v1/responses" \
  -H "Authorization: Bearer $IMOLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "gpt-5.6-terra",
  "input": "Review a migration from a monolith to event-driven services and identify the most important operational trade-offs.",
  "reasoning": {
    "mode": "pro",
    "effort": "medium"
  }
}'