> ## Documentation Index
> Fetch the complete documentation index at: https://docs.requesty.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Create Decision
> Ask typed questions about a piece of content and get a calibrated answer for each one.
```bash cURL theme={"dark"}
curl https://router.requesty.ai/v1/decisions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
-d '{
"model": "openai/gpt-6-luna",
"state": "The dragon heist in this fantasy novel had me hooked, but the ending dragged a little.",
"questions": {
"positive": {
"type": "noul",
"instructions": "The review is positive."
},
"genre": {
"type": "choice",
"instructions": "Which genre is the book?",
"criteria": {
"fantasy": "Magic, myth or invented worlds",
"mystery": "A crime or puzzle to solve",
"romance": "A love story"
}
},
"rating": {
"type": "score",
"instructions": "How much did the reviewer enjoy the book?",
"criteria": ["Hated it", "Disliked it", "Neutral", "Liked it", "Loved it"]
}
}
}'
```
```python Python theme={"dark"}
import httpx
response = httpx.post(
"https://router.requesty.ai/v1/decisions",
headers={"Authorization": "Bearer YOUR_REQUESTY_API_KEY"},
json={
"model": "openai/gpt-6-luna",
"state": "The dragon heist in this fantasy novel had me hooked, but the ending dragged a little.",
"questions": {
"positive": {"type": "noul", "instructions": "The review is positive."},
},
},
)
answer = response.json()["answers"]["positive"]
if answer["type"] == "refusal":
print("The model declined to answer")
else:
print(answer["noul"])
```
```typescript TypeScript theme={"dark"}
const response = await fetch("https://router.requesty.ai/v1/decisions", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.REQUESTY_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
model: "openai/gpt-6-luna",
state: "The dragon heist in this fantasy novel had me hooked, but the ending dragged a little.",
questions: {
positive: { type: "noul", instructions: "The review is positive." },
},
}),
});
const { answers } = await response.json();
const answer = answers.positive;
console.log(answer.type === "refusal" ? "The model declined to answer" : answer.noul);
```
```json Response 200 theme={"dark"}
{
"answers": {
"genre": {
"type": "choice",
"choice": "fantasy",
"confidence": 1,
"probabilities": {
"fantasy": 1,
"mystery": 0,
"romance": 0
}
},
"positive": {
"type": "noul",
"noul": 0.73
},
"rating": {
"type": "score",
"score": 3.1,
"confidence": 0.87,
"probabilities": [
{ "value": 0, "probability": 0.01 },
{ "value": 1, "probability": 0 },
{ "value": 2, "probability": 0 },
{ "value": 3, "probability": 0.86 },
{ "value": 4, "probability": 0.13 }
]
}
},
"id": "rqsty-dcsn-15e2efde-f7c6-49cb-92ae-c763b1b16193",
"model": "gpt-6-luna",
"provider": "openai",
"usage": {
"input_tokens": 424,
"output_tokens": 0
}
}
```
```json Refusal theme={"dark"}
{
"answers": {
"positive": {
"type": "refusal"
}
},
"id": "rqsty-dcsn-75924058-d3b3-471a-b19c-4e003550bcc7",
"model": "gpt-6-luna",
"provider": "openai",
"usage": {
"input_tokens": 5,
"output_tokens": 0
}
}
```
Ask typed questions about a piece of content and get a calibrated answer for each one. List the available models with [`/v1/models/decision`](/api-reference/endpoint/models-decision-list). See [Decisions](/features/decisions) for the question types.
Every answer has a `type`. A model can decline a question and return `{"type": "refusal"}` for it, so check `type` before reading the value. See OpenAI's [Decisions guide](https://developers.openai.com/api/docs/guides/decisions).
Responses are not streamed.
## Images
`cloudflare/clef` and `openai/gpt-6-luna` also look at images. Send them in `images` as base64 data URLs. Remote image URLs are rejected.
```bash theme={"dark"}
curl https://router.requesty.ai/v1/decisions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_REQUESTY_API_KEY" \
-d '{
"model": "openai/gpt-6-luna",
"state": "Listing title: Red leather armchair",
"images": ["data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD..."],
"questions": {
"matches_title": {
"type": "noul",
"instructions": "The photo shows the item in the title."
},
"color": {
"type": "choice",
"instructions": "What is the main colour in the photo?",
"criteria": {
"red": "Red",
"green": "Green",
"blue": "Blue"
}
}
}
}'
```
```json theme={"dark"}
{
"answers": {
"color": {
"type": "choice",
"choice": "red",
"confidence": 0.99,
"probabilities": {
"blue": 0,
"green": 0.01,
"red": 0.99
}
},
"matches_title": {
"type": "noul",
"noul": 0.52
}
},
"id": "rqsty-dcsn-b57050ea-7298-4d59-a0b0-ffc87db60fe8",
"model": "gpt-6-luna",
"provider": "openai",
"usage": {
"input_tokens": 287,
"output_tokens": 0
}
}
```
## OpenAPI
````yaml POST /v1/decisions
openapi: 3.0.3
info:
title: Requesty Inference API
description: >-
Requesty Inference API for AI model routing. OpenAI-compatible endpoints for
chat completions, embeddings, images, audio, and more.
version: 1.0.0
servers:
- url: https://router.requesty.ai
description: Inference router endpoint
security:
- BearerAuth: []
paths:
/v1/decisions:
servers:
- url: https://router.requesty.ai
description: Inference router endpoint
post:
summary: Create decision
description: >-
Ask typed questions about a piece of content and get a calibrated answer
for each one.
operationId: createDecision
requestBody:
required: true
content:
application/json:
schema:
$ref: '#/components/schemas/DecisionRequest'
example:
model: openai/gpt-6-luna
state: >-
The dragon heist in this fantasy novel had me hooked, but the
ending dragged a little.
questions:
positive:
type: noul
instructions: The review is positive.
genre:
type: choice
instructions: Which genre is the book?
criteria:
fantasy: Magic, myth or invented worlds
mystery: A crime or puzzle to solve
romance: A love story
rating:
type: score
instructions: How much did the reviewer enjoy the book?
criteria:
- Hated it
- Disliked it
- Neutral
- Liked it
- Loved it
responses:
'200':
description: Decision response
content:
application/json:
schema:
$ref: '#/components/schemas/DecisionResponse'
example:
answers:
genre:
type: choice
choice: fantasy
confidence: 1
probabilities:
fantasy: 1
mystery: 0
romance: 0
positive:
type: noul
noul: 0.73
rating:
type: score
score: 3.1
confidence: 0.87
probabilities:
- value: 0
probability: 0.01
- value: 1
probability: 0
- value: 2
probability: 0
- value: 3
probability: 0.86
- value: 4
probability: 0.13
id: rqsty-dcsn-15e2efde-f7c6-49cb-92ae-c763b1b16193
model: gpt-6-luna
provider: openai
usage:
input_tokens: 424
output_tokens: 0
'400':
description: Bad request - malformed payload or invalid parameters.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'401':
description: Unauthorized - missing or empty Authorization header.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'402':
description: Payment required - organization balance exhausted.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'403':
description: Forbidden - invalid token or model not in access list.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'404':
description: Not found - provider/model not supported.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'413':
description: Payload too large - the state exceeds the model's context window.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'429':
description: Rate limit exceeded. Retry after the Retry-After header value.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'500':
description: Internal server error.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
'502':
description: Bad gateway - upstream provider returned an invalid response.
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
components:
schemas:
DecisionRequest:
type: object
required:
- model
- state
- questions
properties:
model:
type: string
description: >-
A decision model ID from
[`/v1/models/decision`](/api-reference/endpoint/models-decision-list).
example: openai/gpt-6-luna
state:
description: 'The content to evaluate: a string, an object or an array.'
questions:
type: object
description: Map of question ID to question. Each ID gets one entry in `answers`.
additionalProperties:
$ref: '#/components/schemas/DecisionQuestion'
images:
type: array
items:
type: string
description: >-
Base64 image data URLs (`data:image/...;base64,...`). Only for
models that accept images (`cloudflare/clef`, `openai/gpt-6-luna`).
Remote URLs are rejected.
DecisionResponse:
type: object
required:
- answers
- model
- usage
properties:
answers:
type: object
description: One answer per question ID.
additionalProperties:
$ref: '#/components/schemas/DecisionAnswer'
id:
type: string
example: rqsty-dcsn-15e2efde-f7c6-49cb-92ae-c763b1b16193
model:
type: string
description: The upstream model name.
example: gpt-6-luna
provider:
type: string
example: openai
usage:
$ref: '#/components/schemas/DecisionUsage'
ErrorResponse:
type: object
required:
- error
properties:
error:
type: object
required:
- origin
- message
properties:
origin:
type: string
enum:
- router
- provider
description: >-
Whether the error originated from Requesty's router or an
upstream provider.
message:
type: string
description: Human-readable error description.
DecisionQuestion:
type: object
required:
- type
properties:
type:
type: string
enum:
- noul
- choice
- score
description: >-
`noul`: yes/no, answered with a probability. `choice`: pick one
option from `criteria`. `score`: a position on the ordered scale in
`criteria`.
instructions:
description: What to ask about the state. Usually a string.
oneOf:
- type: string
- type: array
items:
type: string
criteria:
description: >-
For `choice`: an object mapping each option to its description. For
`score`: a list of labels from lowest to highest. Not used by
`noul`.
oneOf:
- type: object
additionalProperties:
type: string
nullable: true
- type: array
items:
type: string
DecisionAnswer:
type: object
required:
- type
description: >-
Answer to one question. Check `type` first: a model can return `refusal`
instead of answering.
properties:
type:
type: string
enum:
- noul
- choice
- score
- refusal
noul:
type: number
description: '`noul` only. Probability between 0 and 1 that the statement is true.'
choice:
type: string
description: '`choice` only. The most likely option.'
score:
type: number
description: >-
`score` only. Probability-weighted average of the zero-based label
indices, so it can be fractional.
confidence:
type: number
description: >-
`choice` and `score` only. Confidence in the answer, between 0 and
1.
probabilities:
description: >-
`choice` and `score` only. Probability per option (`choice`) or per
label index (`score`). Depending on the model, score probabilities
are an object keyed by index or a list of `{value, probability}`.
oneOf:
- type: object
additionalProperties:
type: number
- type: array
items:
type: object
properties:
value:
type: integer
probability:
type: number
legend:
type: object
additionalProperties:
type: string
description: '`score` only, on some models. Maps each label index to its label.'
DecisionUsage:
type: object
required:
- input_tokens
- output_tokens
properties:
input_tokens:
type: integer
output_tokens:
type: integer
description: Can be 0.
securitySchemes:
BearerAuth:
type: http
scheme: bearer
description: API key for authentication
````
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.