> ## Documentation Index
> Fetch the complete documentation index at: https://evolink.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# GPT Decisions - Complete Parameter Reference
> - GPT-6 Luna Decisions (`gpt-6-luna-decisions`) evaluates text, images, or both and returns three types of structured answers: `predicate` returns the probability of “yes”, `choice` selects one of the supplied options, and `score` calculates a probability-weighted score over ordered levels
- Synchronous API: all answers are returned in a single request; no streaming or task polling
- The request body matches the OpenAI Decisions API: with the OpenAI SDK, point `base_url` to EvoLink. This endpoint also accepts the OpenAI model ID `gpt-6-luna`
- Only input tokens are billed: output, cache reads, and cache writes are free. A minimum charge applies to each request; when a single input exceeds 272,000 tokens, the input price for the entire request is 2 times the standard rate. See [Model pricing](https://evolink.ai/pricing) for current prices
- Images must be inline base64 data URLs (`data:image/...;base64,...`); image URLs and `file_id` are not supported. Up to 128 images per request
- This endpoint rejects unknown fields: do not send Chat Completions parameters such as `stream`, `temperature`, or `max_tokens`
- Suitable for content classification, ticket routing, relevance checks, and rule-based scoring. For text generation or custom JSON structures, use the [Responses API](/en/api-manual/language-series/gpt/responses/responses-reference)
**BaseURL**: The default BaseURL is `https://direct.evolink.ai`, which has better support for text models and long-lived connections. `https://api.evolink.ai` is the primary endpoint for multimodal services and serves as a fallback address for text models.
**Model ID is exclusive to this endpoint**: `gpt-6-luna-decisions` can only be used on `/v1/decisions`. Using it on Chat Completions, Responses, or other endpoints returns `400 model_endpoint_mismatch`. For conversation or text generation, use `gpt-6-luna` with [Chat Completions](/docs/en/api-manual/language-series/gpt/chat-completions/chat-completions-reference) or the [Responses API](/docs/en/api-manual/language-series/gpt/responses/responses-reference).
**Choosing a question type**
| Type | Suitable questions | How to use the result |
| - | - | - |
| `predicate` | Yes/no questions, such as whether a message is a complaint or an image shows damage | `probability` is the probability of “yes”; set your own threshold and treat values above it as a match |
| `choice` | Select one of several mutually exclusive categories, such as the department to route a ticket to | `choice` is the selected option; consider human review when `confidence` is low |
| `score` | Evaluate ordered levels, such as issue severity | `score` is the probability-weighted level score (indices start at 0) and may be fractional |
You can include multiple independent questions in one request; they share the same `input`. If a later question depends on an earlier answer, split them into two requests.
**Difference from Chat Completions / Responses**: This endpoint only accepts `model`, `input`, `questions`, and `safety_identifier`. Any additional field returns `400 unknown_parameter`; images must be base64 data URLs; streaming is not supported.
## OpenAPI
````yaml en/api-manual/language-series/gpt/decisions/decisions-reference.json POST /v1/decisions
openapi: 3.1.0
info:
title: GPT Decisions - Complete Parameter Reference
description: >-
Use the OpenAI-compatible Decisions API to have GPT-6 Luna make structured
decisions about text and images: yes/no probabilities, a single choice, or
scoring on ordered levels.
license:
name: MIT
version: 1.0.0
servers:
- url: https://direct.evolink.ai
description: Production (recommended)
- url: https://api.evolink.ai
description: Fallback endpoint
security:
- bearerAuth: []
tags:
- name: Decisions
description: 'Structured decisions: classification, routing, and scoring'
paths:
/v1/decisions:
post:
tags:
- Decisions
summary: GPT-6 Luna Decisions (structured decisions)
description: >-
- GPT-6 Luna Decisions (`gpt-6-luna-decisions`) evaluates text, images,
or both and returns three types of structured answers: `predicate`
returns the probability of “yes”, `choice` selects one of the supplied
options, and `score` calculates a probability-weighted score over
ordered levels
- Synchronous API: all answers are returned in a single request; no
streaming or task polling
- The request body matches the OpenAI Decisions API: with the OpenAI
SDK, point `base_url` to EvoLink. This endpoint also accepts the OpenAI
model ID `gpt-6-luna`
- Only input tokens are billed: output, cache reads, and cache writes
are free. A minimum charge applies to each request; when a single input
exceeds 272,000 tokens, the input price for the entire request is 2
times the standard rate. See [Model pricing](https://evolink.ai/pricing)
for current prices
- Images must be inline base64 data URLs (`data:image/...;base64,...`);
image URLs and `file_id` are not supported. Up to 128 images per request
- This endpoint rejects unknown fields: do not send Chat Completions
parameters such as `stream`, `temperature`, or `max_tokens`
- Suitable for content classification, ticket routing, relevance checks,
and rule-based scoring. For text generation or custom JSON structures,
use the [Responses
API](/en/api-manual/language-series/gpt/responses/responses-reference)
operationId: createDecision
requestBody:
required: true
content:
application/json:
schema:
$ref: '#/components/schemas/DecisionRequest'
examples:
predicate:
summary: Yes/no decision
value:
model: gpt-6-luna-decisions
input: The package arrived with a broken screen.
questions:
- type: predicate
name: damaged
instructions: Does the customer report a damaged item?
choice:
summary: Single choice (routing)
value:
model: gpt-6-luna-decisions
input: I was charged twice for my order.
questions:
- type: choice
name: department
instructions: Which department should handle this complaint?
choices:
- value: billing
description: Payments, invoices, and refunds.
- value: technical
description: Problems using the product.
- value: other
description: Requests outside these categories.
score:
summary: Scoring on ordered levels
value:
model: gpt-6-luna-decisions
input: Export fails in Safari but works in Chrome.
questions:
- type: score
name: severity
instructions: How severe is this issue?
levels:
- label: Cosmetic
description: Appearance only; no lost functionality.
- label: Workaround available
description: A task fails, but another way works.
- label: Fully blocked
description: A task fails with no workaround.
multiple:
summary: Ask multiple questions in one request
value:
model: gpt-6-luna-decisions
input: >-
I was charged twice for my order and the app crashes when I
open the invoice.
questions:
- type: predicate
name: is_billing
instructions: Is this about a billing problem?
- type: choice
name: department
instructions: Which department should handle this complaint?
choices:
- value: billing
description: Payments, invoices, and refunds.
- value: technical
description: Problems using the product.
- value: other
description: Requests outside these categories.
- type: score
name: severity
instructions: How severe is this issue?
levels:
- label: Cosmetic
description: Appearance only.
- label: Workaround available
description: A task fails, but another way works.
- label: Fully blocked
description: A task fails with no workaround.
image:
summary: Evaluate an image
value:
model: gpt-6-luna-decisions
input:
- role: user
content:
- type: input_text
text: Look at the image.
- type: input_image
image_url: >-
data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAACAAAAAgCAIAAAD8GO2jAAAAKklEQVR42mM4ISdHU8QwasGoBaMWjFowasGoBaMWjFowasGoBaMWDBULACXLED1gHZEpAAAAAElFTkSuQmCC
questions:
- type: predicate
name: is_red
instructions: Is the image mostly red?
responses:
'200':
description: >-
One answer per question, in the same order as the questions in the
request
content:
application/json:
schema:
$ref: '#/components/schemas/DecisionResponse'
example:
model: gpt-6-luna
answers:
- type: predicate
name: is_billing
probability: 1
- type: choice
name: department
choice: billing
probabilities:
- value: billing
probability: 0.98
- value: technical
probability: 0.01
- value: other
probability: 0.01
confidence: 0.97
- type: score
name: severity
score: 1.34
probabilities:
- value: 0
label: Cosmetic
probability: 0.04
- value: 1
label: Workaround available
probability: 0.58
- value: 2
label: Fully blocked
probability: 0.38
confidence: 0.37
usage:
input_tokens: 426
input_tokens_details:
cached_tokens: 0
cache_write_tokens: 0
output_tokens: 0
output_tokens_details:
reasoning_tokens: 0
total_tokens: 426
'400':
description: >-
Invalid request parameters. `param` identifies the field with the
error. Common causes: missing `questions`, unknown fields, images
that are not base64 data URLs, a message `role` other than `user`,
or duplicate question `name` values. Using this model ID on another
endpoint also returns 400 (`model_endpoint_mismatch`). Requests that
fail validation are not billed
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
examples:
unknown_parameter:
summary: Unknown field supplied
value:
error:
message: >-
Unknown parameter: 'stream'. (request id:
20261007223656995142218QgCDR3TF)
type: invalid_request_error
param: stream
code: unknown_parameter
image_not_data_url:
summary: Image supplied as a URL
value:
error:
message: >-
Invalid 'input[0].content[1].image_url': string does not
match pattern. Expected a string that matches the
pattern '^data:'. (request id:
2026100722365642880926evUSStdF)
type: invalid_request_error
param: input[0].content[1].image_url
code: invalid_value
missing_questions:
summary: Missing questions
value:
error:
message: >-
Missing required parameter: 'questions'. (request id:
20261007223656416678385UfxZWwZD)
type: invalid_request_error
param: questions
code: missing_required_parameter
'401':
description: 'Unauthorized: API Key is missing or invalid'
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: unauthorized
message: >-
API key is required (request id:
202610072235193122631737Nn7cfZ5)
param: null
type: authentication_error
'402':
description: Insufficient balance
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'403':
description: >-
This API Key has a model allowlist that includes neither
`gpt-6-luna-decisions` nor `gpt-6-luna`
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'404':
description: >-
This model is not enabled for the Decisions API. Currently only
`gpt-6-luna-decisions` (or `gpt-6-luna`) is supported
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: model_not_found
message: >-
Model 'gpt-5.5' is not available for the Decisions API (POST
/v1/decisions) with this API key. This error is permanent —
do not retry with the same model id. Call GET /v1/models:
models that support this endpoint are listed with the
'-decisions' suffix. (request id:
20261007223657322725427OrY2GeLp)
param: null
retryable: false
type: invalid_request_error
'429':
description: Rate limit exceeded; retry later
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'500':
description: Internal server error
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'503':
description: Service temporarily unavailable; retry later
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
components:
schemas:
DecisionRequest:
type: object
additionalProperties: false
required:
- model
- input
- questions
properties:
model:
type: string
description: >-
Model ID. Use `gpt-6-luna-decisions`; this endpoint also accepts the
OpenAI model ID `gpt-6-luna`. Both have identical behavior and
billing; usage and billing records are attributed to
`gpt-6-luna-decisions`.
`gpt-6-luna-decisions` is only available on this endpoint. Using it
on Chat Completions, Responses, or other endpoints returns `400
model_endpoint_mismatch`.
enum:
- gpt-6-luna-decisions
- gpt-6-luna
example: gpt-6-luna-decisions
input:
description: >-
The shared basis for all decisions: a text string or an array of
user messages, which can contain text and inline images.
oneOf:
- type: string
example: I was charged twice for my order.
- type: array
minItems: 1
items:
$ref: '#/components/schemas/DecisionInputMessage'
questions:
type: array
minItems: 1
description: >-
Questions to evaluate; at least 1 is required. Questions are
independent and can mix different types; question `name` values must
be unique within a request.
| Type | Purpose | Main result |
|---|---|---|
| `predicate` | Check whether a condition holds | `probability`:
probability that it holds (0 to 1) |
| `choice` | Select one of the supplied options | `choice`: the
selected option |
| `score` | Score on ordered levels | `score`: probability-weighted
level score |
If a later question depends on an earlier answer, split them into
two requests.
items:
oneOf:
- $ref: '#/components/schemas/PredicateQuestion'
- $ref: '#/components/schemas/ChoiceQuestion'
- $ref: '#/components/schemas/ScoreQuestion'
safety_identifier:
type: string
description: >-
Optional. Your own end-user identifier (an opaque string), forwarded
unchanged to the model provider for abuse detection
DecisionResponse:
type: object
properties:
model:
type: string
description: The model that actually performed the decisions
example: gpt-6-luna
answers:
type: array
description: >-
One answer per question, in the same order as the questions in the
request. Use `name` to match the question and `type` to distinguish
answer types
items:
oneOf:
- $ref: '#/components/schemas/PredicateAnswer'
- $ref: '#/components/schemas/ChoiceAnswer'
- $ref: '#/components/schemas/ScoreAnswer'
- $ref: '#/components/schemas/RefusalAnswer'
usage:
$ref: '#/components/schemas/DecisionUsage'
ErrorResponse:
type: object
properties:
error:
type: object
properties:
message:
type: string
description: >-
Error description, ending with `(request id: ...)`. Provide this
ID when troubleshooting
type:
type: string
description: Error type
param:
type:
- string
- 'null'
description: Path to the invalid parameter
code:
type:
- string
- 'null'
description: Error code
DecisionInputMessage:
type: object
required:
- role
- content
properties:
role:
type: string
enum:
- user
description: Only `user` is accepted
content:
description: A text string or an array of text and image blocks
oneOf:
- type: string
- type: array
items:
oneOf:
- $ref: '#/components/schemas/InputTextPart'
- $ref: '#/components/schemas/InputImagePart'
PredicateQuestion:
type: object
required:
- type
- instructions
properties:
type:
type: string
enum:
- predicate
name:
type: string
description: The question name, returned unchanged in the corresponding answer
instructions:
type: string
description: The condition to evaluate, written as a yes/no question
ChoiceQuestion:
type: object
required:
- type
- instructions
- choices
properties:
type:
type: string
enum:
- choice
name:
type: string
description: The question name, returned unchanged in the corresponding answer
instructions:
type: string
description: The question to answer
choices:
type: array
minItems: 1
description: >-
Available options. Their meanings should be mutually exclusive; if
the categories may not cover every case, add a fallback such as
`other`
items:
type: object
required:
- value
properties:
value:
description: >-
The value returned when this option is selected. Strings and
booleans are different types: `true` and `"true"` count as two
options
oneOf:
- type: string
- type: boolean
description:
type: string
description: When to select this option
ScoreQuestion:
type: object
required:
- type
- instructions
- levels
properties:
type:
type: string
enum:
- score
name:
type: string
description: The question name, returned unchanged in the corresponding answer
instructions:
type: string
description: The question to answer
levels:
type: array
minItems: 1
description: Levels, ordered from lowest to highest; level indices start at 0
items:
type: object
required:
- label
properties:
label:
type: string
description: Level name
description:
type: string
description: Criteria for reaching this level
PredicateAnswer:
type: object
properties:
type:
type: string
enum:
- predicate
name:
type:
- string
- 'null'
description: >-
The name assigned to this question in the request; `null` if no name
was supplied
probability:
type: number
description: >-
Estimated probability that the condition holds, from 0 to 1. Set a
decision threshold using your own business data
ChoiceAnswer:
type: object
properties:
type:
type: string
enum:
- choice
name:
type:
- string
- 'null'
description: >-
The name assigned to this question in the request; `null` if no name
was supplied
choice:
description: The selected option, with the same type as `value` in the request
oneOf:
- type: string
- type: boolean
probabilities:
type: array
description: Probability for each option
items:
type: object
properties:
value:
oneOf:
- type: string
- type: boolean
probability:
type: number
confidence:
type: number
description: Confidence in this choice, from 0 to 1
ScoreAnswer:
type: object
properties:
type:
type: string
enum:
- score
name:
type:
- string
- 'null'
description: >-
The name assigned to this question in the request; `null` if no name
was supplied
score:
type: number
description: >-
The probability-weighted average of the level indices, so it can
fall between two levels. For example, probabilities of 0.04 / 0.58 /
0.38 across three levels yield a score of 1.34
probabilities:
type: array
description: Probability for each level
items:
type: object
properties:
value:
type: integer
description: Level index, starting at 0
label:
type: string
probability:
type: number
confidence:
type: number
description: Confidence in this score, from 0 to 1
RefusalAnswer:
type: object
description: >-
The model refused to answer this question. Answers to other questions
are unaffected; the request still returns 200 and is billed as usual
properties:
type:
type: string
enum:
- refusal
name:
type:
- string
- 'null'
description: >-
The name assigned to this question in the request; `null` if no name
was supplied
DecisionUsage:
type: object
description: >-
Token usage. Only input tokens are billed on this endpoint: output,
cache reads, and cache writes are free. A minimum charge applies to each
request; when a single input exceeds 272,000 tokens, the input price for
the entire request is 2 times the standard rate. See [Model
pricing](https://evolink.ai/pricing) for current prices.
properties:
input_tokens:
type: integer
description: >-
Billable input tokens, including text, images, and fixed overhead
per request and per question. Reference values: one sentence with
one `predicate` question uses about 160 tokens; a 1536×1024 image
uses about 1,540 tokens
example: 426
input_tokens_details:
type: object
description: >-
Input token details. Caching is not billed separately on this
endpoint
properties:
cached_tokens:
type: integer
example: 0
cache_write_tokens:
type: integer
example: 0
output_tokens:
type: integer
description: Output is not billed on this endpoint
example: 0
output_tokens_details:
type: object
properties:
reasoning_tokens:
type: integer
example: 0
total_tokens:
type: integer
description: Total tokens
example: 426
InputTextPart:
type: object
required:
- type
- text
properties:
type:
type: string
enum:
- input_text
text:
type: string
description: Text content
InputImagePart:
type: object
required:
- type
- image_url
properties:
type:
type: string
enum:
- input_image
image_url:
type: string
pattern: '^data:'
description: >-
An inline base64 data URL, such as `data:image/png;base64,...`.
`http(s)` image URLs are rejected; up to 128 images per request
securitySchemes:
bearerAuth:
type: http
scheme: bearer
description: >-
##All endpoints require Bearer Token authentication##
**Get an API Key:**
Visit the [API Key management page](https://evolink.ai/dashboard/keys)
to get your API Key
**Add it to the request header:**
```
Authorization: Bearer YOUR_API_KEY
```
````
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.