Jev
curl --request POST \
--url https://api.gravitex.ai/v1/systemone \
--header 'Authorization: <authorization>' \
--header 'Content-Type: <content-type>'import requests
url = "https://api.gravitex.ai/v1/systemone"
headers = {
"Authorization": "<authorization>",
"Content-Type": "<content-type>"
}
response = requests.post(url, headers=headers)
print(response.text)const options = {
method: 'POST',
headers: {Authorization: '<authorization>', 'Content-Type': '<content-type>'}
};
fetch('https://api.gravitex.ai/v1/systemone', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));<?php
$curl = curl_init();
curl_setopt_array($curl, [
CURLOPT_URL => "https://api.gravitex.ai/v1/systemone",
CURLOPT_RETURNTRANSFER => true,
CURLOPT_ENCODING => "",
CURLOPT_MAXREDIRS => 10,
CURLOPT_TIMEOUT => 30,
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
CURLOPT_CUSTOMREQUEST => "POST",
CURLOPT_HTTPHEADER => [
"Authorization: <authorization>",
"Content-Type: <content-type>"
],
]);
$response = curl_exec($curl);
$err = curl_error($curl);
curl_close($curl);
if ($err) {
echo "cURL Error #:" . $err;
} else {
echo $response;
}package main
import (
"fmt"
"net/http"
"io"
)
func main() {
url := "https://api.gravitex.ai/v1/systemone"
req, _ := http.NewRequest("POST", url, nil)
req.Header.Add("Authorization", "<authorization>")
req.Header.Add("Content-Type", "<content-type>")
res, _ := http.DefaultClient.Do(req)
defer res.Body.Close()
body, _ := io.ReadAll(res.Body)
fmt.Println(string(body))
}HttpResponse<String> response = Unirest.post("https://api.gravitex.ai/v1/systemone")
.header("Authorization", "<authorization>")
.header("Content-Type", "<content-type>")
.asString();require 'uri'
require 'net/http'
url = URI("https://api.gravitex.ai/v1/systemone")
http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true
request = Net::HTTP::Post.new(url)
request["Authorization"] = '<authorization>'
request["Content-Type"] = '<content-type>'
response = http.request(request)
puts response.read_bodyDecision models
Jev
TypeSafe Jev decision model: return choices, scores, yes/no probabilities, or image bounding boxes with confidence
POST
/
v1
/
systemone
Jev
curl --request POST \
--url https://api.gravitex.ai/v1/systemone \
--header 'Authorization: <authorization>' \
--header 'Content-Type: <content-type>'import requests
url = "https://api.gravitex.ai/v1/systemone"
headers = {
"Authorization": "<authorization>",
"Content-Type": "<content-type>"
}
response = requests.post(url, headers=headers)
print(response.text)const options = {
method: 'POST',
headers: {Authorization: '<authorization>', 'Content-Type': '<content-type>'}
};
fetch('https://api.gravitex.ai/v1/systemone', options)
.then(res => res.json())
.then(res => console.log(res))
.catch(err => console.error(err));<?php
$curl = curl_init();
curl_setopt_array($curl, [
CURLOPT_URL => "https://api.gravitex.ai/v1/systemone",
CURLOPT_RETURNTRANSFER => true,
CURLOPT_ENCODING => "",
CURLOPT_MAXREDIRS => 10,
CURLOPT_TIMEOUT => 30,
CURLOPT_HTTP_VERSION => CURL_HTTP_VERSION_1_1,
CURLOPT_CUSTOMREQUEST => "POST",
CURLOPT_HTTPHEADER => [
"Authorization: <authorization>",
"Content-Type: <content-type>"
],
]);
$response = curl_exec($curl);
$err = curl_error($curl);
curl_close($curl);
if ($err) {
echo "cURL Error #:" . $err;
} else {
echo $response;
}package main
import (
"fmt"
"net/http"
"io"
)
func main() {
url := "https://api.gravitex.ai/v1/systemone"
req, _ := http.NewRequest("POST", url, nil)
req.Header.Add("Authorization", "<authorization>")
req.Header.Add("Content-Type", "<content-type>")
res, _ := http.DefaultClient.Do(req)
defer res.Body.Close()
body, _ := io.ReadAll(res.Body)
fmt.Println(string(body))
}HttpResponse<String> response = Unirest.post("https://api.gravitex.ai/v1/systemone")
.header("Authorization", "<authorization>")
.header("Content-Type", "<content-type>")
.asString();require 'uri'
require 'net/http'
url = URI("https://api.gravitex.ai/v1/systemone")
http = Net::HTTP.new(url.host, url.port)
http.use_ssl = true
request = Net::HTTP::Post.new(url)
request["Authorization"] = '<authorization>'
request["Content-Type"] = '<content-type>'
response = http.request(request)
puts response.read_bodyIntroduction
Jev is TypeSafeโs System One decision model. It does not generate text. Instead, it answers questions with a defined answer space (questions) about the content you provide (state), returning structured results your code can use directly:
- Choice (
choice): Select one of your options, such as which department should handle a support ticket. - Score (
score): Rate content against your levels, such as customer anger: 0 calm / 1 dissatisfied / 2 angry. - Yes/no (
noul): Return the probability of yes, such as whether the customer is requesting a refund. - Bounding box (
box): Locate a described target in an image and return its four edges, such as scratches on a car.
The request structure and successful response fields follow TypeSafeโs System One API. The platform may wrap error responses; see Error handling.
Jev does not support conversational endpoints such as OpenAIโs
/v1/chat/completions or /v1/responses. Call it through /v1/systemone below. It does not write replies, generate code, or explain its reasoning.Authentication
application/json, or multipart/form-data for multipart image uploads. See Multipart image uploads.Set to
true. The official specification requires this header for requests containing images (assets) or Box questions (box).Set to
true to enable long context and raise the context limit from 64k to 262k. See Context length.Model information
| Model name | Description | Question types | Input |
|---|---|---|---|
jev-latest | Latest stable release; currently points to jev-1.14.0 | choice / score / noul / box | Text, images |
jev-preview | Latest release, including previews; currently also points to jev-1.14.0 | choice / score / noul / box | Text, images |
jev-1.14.0 | Pinned Jev 1.14 version | choice / score / noul / box | Text, images |
jev-1.13.0 | Pinned Jev 1.13 version | choice / score / noul (no box support) | Text, images |
jev-latestandjev-previeware aliases that move when TypeSafe releases new versions, so answers may change. The responseโsmodelfield identifies the version that answered.- If you have tuned confidence thresholds for a version, use its pinned model ID and switch only after evaluating the new version.
| Item | Description |
|---|---|
| Endpoint type | Synchronous response |
| Streaming | Not supported |
| Questions per request | No separate limit, provided the request fits the context limits |
| Image formats | PNG, JPEG |
| Billing | Input tokens only; output tokens are free |
Endpoint
POST https://api.gravitex.ai/v1/systemone
Request parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
model | string | Yes | Model name; see Model information |
state | string / object / array | Yes | Content all questions evaluate; see state |
questions | object | Yes | At least one named question; keys are your question names and values are question definitions. See questions |
assets | object | No | Images keyed by placeholders, with image definitions as values. See assets |
Jev has no sampling parameters such as
temperature or max_tokens, and no output length setting is needed.state
Provide the content to evaluate in any of three forms:- String: Text such as a customer message, support ticket, or complete contract.
- Object: Structured data such as
{"subject": "Duplicate charge", "message": "Please help"}. Use it to combine messages, order records, and business rules. - Array: Multiple pieces of text, such as a conversation history.
"state": {
"customer_message": "My order was charged twice, and nobody has addressed it for three days",
"orders": [{ "id": "A1024", "charged_times": 2 }],
"refund_policy": "Duplicate charges are eligible for a full refund of the extra payment"
}
questions
A collection of named questions. The response returns answers under the same names inanswers. Each value defines a question:
| Field | Type | Required | Description |
|---|---|---|---|
type | string | Yes | Question type: choice, score, noul, or box |
instructions | string / object / array | No | What the model should evaluate |
criteria | object / array | Depends on type | Options or scoring levels; see the table below |
| Question type | criteria format | Requirements |
|---|---|---|
choice | Object: {option_name: option_description} | Required, 1โ255 options. A description may be empty; the option is then interpreted by name alone. |
score | Array of level descriptions, ordered from low to high | Required, 1โ10 levels; at least 2 recommended. The first level is 0, increasing by 1. |
noul | Object: {"true": "What counts as yes", "false": "What counts as no"} | Optional; provide at least instructions or criteria, otherwise the upstream service returns an error. |
box | Omit | Only jev-1.14.0 and aliases pointing to it support Box. An image and TypeSafe-Danger-Zone-Images: true are required. |
"questions": {
"department": {
"type": "choice",
"instructions": "Which department should handle this ticket?",
"criteria": {
"billing": "Payments, charges, refunds",
"technical": "Feature failures, errors",
"sales": "Presales questions, plan upgrades"
}
},
"anger": {
"type": "score",
"instructions": "How angry is the customer?",
"criteria": ["Calm", "Somewhat dissatisfied", "Very angry"]
},
"wants_refund": {
"type": "noul",
"instructions": "Is the customer requesting a refund?",
"criteria": { "true": "Explicitly requests money back", "false": "Does not request a refund" }
}
}
- Questions are independent. Each is evaluated against
statewithout seeing the other answers. - Specific option and level descriptions make decisions more consistent. Avoid overlapping options.
assets
A collection of images keyed by your placeholders (a new UUID per image is recommended). A placeholder anywhere in a string instate, question instructions, or criteria is replaced with the corresponding image.
| Field | Type | Required | Description |
|---|---|---|---|
type | string | No | Must be image when supplied; explicitly included in the examples |
format | string | Required with data_base64 | png, jpg, or jpeg; the actual format is detected from the image bytes |
data_base64 | string | One of two methods | Base64 image data without a data:image/...;base64, prefix |
part | string | One of two methods | Name of the image part in a multipart request. When set, data_base64 is ignored. |
"state": "photo",
"assets": {
"photo": {
"type": "image",
"format": "png",
"data_base64": "iVBORw0KGgoAAAANSUhEUgAAARgAAAEYCAIAAAAI7H7bAAAC/UlEQVR42u3TwQ0AMAgDsey/NN0BPlT4lAEqVKckjYsTSCBJIEkgSQJJAkkCSQJJEkgSSBJIEkiSQJJAkkCSQJIEkgSSBJIEkiSQJJAkkCSQJIEkgSSBJAkkCSQJJAkkSSBJIEkgSSBJAkkCSQJJAkkSSBJIEkgSSJJAkkCSQJJAkgSSBJIEkgSSJJAkkCSQJIEkgSSBJIEkCSQJJAkkCSRJIEkgSSBJIEkCSQJJAkkCSRJIEkgSSBJIkkCSQJJAkkCSBJIEkgSSJJAkkCSQmi+NXRxIIBlIIBlIIBlIIIFkIIFkIIFkIIFkIIEEkoEEkoEEkoEEkoEEEkgGEkgGEkgGEkgGEkggGUggGUggGUggGUgggWQggWQggWQggQQSSCCBZCCBZCCBZCCBBJKBBJKBBJKBBJKBBBJIBhJIBhJIBhJIBhJIIBlIIBlIIBlIIBlIIIFkIIFkIIFkIIFkIIEEkoEEkoEEkoEEkoEEEkgGEkgGEkgGEkgggQQSSAYSSAYSSAYSSCAZSCAZSCAZSCAZSCCBZCCBZCCBZCCBZCCBBJKBBJKBBJKBBJKBBBJIBhJIBhJIBhJIBhJIIBlIIBlIIBlIIPlSIIEEkoEEkoEEkoEEEkgggQSSgQSSgQSSgQQSSAYSSAYSSAYSSAYSSCAZSCAZSCAZSCAZSCCBZCCBZCCBZCCBZCCBBJKBBJKBBJKBBJKBBBJIBhJIBhJIBhJIIIEEEkgGEkgGEkgGEkggGUggGUggGUggGUgggWQggWQggWQggWQggSR9HUgSSBJIEkiSQJJAkkCSQJIEkgSSBJIEkiSQJJAkkCSQJIEkgSSBJIEkCSQJJAkkCSRJIEkgSSBJAkkCSQJJAkkSSBJIEkgSSJJAkkCSQJJAkgSSBJIEkgSSJJAkkCSQJJAkgSSBJIEkgSQJJAkkCSRJIEkgSSBJIEkCSQJJAkkCSRJIEkgSSBJIkkCSQJJAkkCSBJIEkgSSBJIkkCSQJJAkkJxAAkkCSQJJEkgSSNKyHikG6UERft5KAAAAAElFTkSuQmCC"
}
}
The official specification requires
TypeSafe-Danger-Zone-Images: true for requests with assets. Images and Box questions are early features; request formats, answers, and limits may change between versions.Image limits
Images must satisfy all of the following official limits. Images outside the dimension limits are not resized automatically; the whole request fails:| Item | Limit |
|---|---|
| Format | Still PNG or JPEG; animated PNG is not supported |
| Width and height | 280โ1568 pixels each, inclusive |
| Total pixels | 78,400โ2,458,624 |
| Single file | At most 16 MiB; Base64 is measured after decoding |
| Request body | At most 32 MiB, including JSON, Base64, or multipart overhead |
| All images / expanded request | Each at most 32 MiB; repeated image references count again toward the expanded size |
state, instructions, or criteria. Undersized images, invalid formats, and unreferenced assets typically return 400. Exceeding file, dimension, or request size limits may return 413.
Multipart image uploads
For large or multiple images, upload the original files withmultipart/form-data instead of Base64:
- The first part must be named
requestand contain the JSON request body (Content-Type: application/json). - Each image follows in its own part. Part names must be unique, nonempty, and referenced by an assetโs
partfield.
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'TypeSafe-Danger-Zone-Images: true' \
--form 'request={"model":"jev-latest","state":"photo","questions":{"damage":{"type":"noul","instructions":"Are there visible scratches on the car?"}},"assets":{"photo":{"type":"image","part":"img1"}}};type=application/json' \
--form 'img1=@./car.jpg;type=image/jpeg'
Context length
Every request must fit both limits:| Limit | Default | With long context |
|---|---|---|
Per request (state + all questions) | 64k tokens | 262k tokens |
Per question (state + that question) | 32k tokens | 262k tokens |
- Exceeding the limit returns an upstream 400 error with
max_tokens_exceeded. - For long contracts or conversation histories, send
TypeSafe-Danger-Zone-Long-Context: true. Long context has the same per-token price, but is slower and slightly less accurate. Enable it only when needed.
Usage examples
- Choice
- Multiple questions
- Structured state
- Image evaluation
- Bounding box
- Long context
- Python
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "jev-latest",
"state": "Help! My payouts have been failing for 3 days.",
"questions": {
"department": {
"type": "choice",
"instructions": "Which team should handle this?",
"criteria": {
"billing": "Payments, invoicing, refunds",
"technical": "Bugs, outages, integrations",
"sales": "Pricing, upgrades, new accounts"
}
}
}
}'
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "jev-latest",
"state": "My order was charged twice and nobody has addressed it for three days. Refund me or I will file a complaint!",
"questions": {
"department": {
"type": "choice",
"instructions": "Which department should handle this ticket?",
"criteria": { "billing": "Payments, charges, refunds", "technical": "Feature failures, errors", "sales": "Presales questions, plan upgrades" }
},
"anger": {
"type": "score",
"instructions": "How angry is the customer?",
"criteria": ["Calm", "Somewhat dissatisfied", "Very angry"]
},
"wants_refund": { "type": "noul", "instructions": "Is the customer requesting a refund?" }
}
}'
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "jev-1.14.0",
"state": {
"customer_message": "My order was charged twice",
"orders": [{ "id": "A1024", "charged_times": 2 }],
"refund_policy": "Duplicate charges are eligible for a full refund of the extra payment"
},
"questions": {
"refund_allowed": { "type": "noul", "instructions": "Under the refund policy, should this order receive a refund?" }
}
}'
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--header 'TypeSafe-Danger-Zone-Images: true' \
--data '{
"model": "jev-latest",
"state": "photo",
"questions": {
"has_red": { "type": "noul", "instructions": "Does the image contain something red?" }
},
"assets": {
"photo": {
"type": "image",
"format": "png",
"data_base64": "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"
}
}
}'
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--header 'TypeSafe-Danger-Zone-Images: true' \
--data '{
"model": "jev-1.14.0",
"state": "photo",
"questions": {
"red_rectangle": { "type": "box", "instructions": "Locate the red rectangle." },
"has_rectangle": { "type": "noul", "instructions": "Is there a red rectangle in the image?" }
},
"assets": {
"photo": {
"type": "image",
"format": "png",
"data_base64": "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"
}
}
}'
curl --location 'https://api.gravitex.ai/v1/systemone' \
--header 'Authorization: Bearer sk-xxxxxxxxxx' \
--header 'Content-Type: application/json' \
--header 'TypeSafe-Danger-Zone-Long-Context: true' \
--data '{
"model": "jev-latest",
"state": "<Full text of a long contract>",
"questions": {
"has_termination_clause": {
"type": "noul",
"instructions": "Does the contract include a clause allowing either party to terminate the agreement early?"
}
}
}'
import requests
# Install the dependency first: pip install requests
resp = requests.post(
"https://api.gravitex.ai/v1/systemone",
headers={"Authorization": "Bearer sk-xxxxxxxxxx"},
json={
"model": "jev-latest",
"state": "My order was charged twice. Refund me or I will file a complaint!",
"questions": {
"wants_refund": {"type": "noul", "instructions": "Is the customer requesting a refund?"},
},
},
timeout=60,
)
resp.raise_for_status()
answer = resp.json()["answers"]["wants_refund"]
if answer["noul"] >= 0.8:
print("Route to the refund workflow")
Successful response
{
"model": "jev-1.14.0",
"answers": {
"department": {
"type": "choice",
"choice": "billing",
"confidence": 1,
"probabilities": { "billing": 1, "technical": 0, "sales": 0 }
},
"anger": {
"type": "score",
"score": 1.87,
"confidence": 0.8,
"legend": { "0": "Calm", "1": "Somewhat dissatisfied", "2": "Very angry" },
"probabilities": { "0": 0, "1": 0.13, "2": 0.87 }
},
"wants_refund": {
"type": "noul",
"noul": 0.97
}
},
"usage": {
"input_tokens": 443,
"output_tokens": 71
}
}
{
"type": "box",
"box": { "top": 0.2, "bottom": 0.66, "left": 0.19, "right": 0.68 },
"confidence": 0.9
}
Response fields
| Field | Type | Description |
|---|---|---|
model | string | Version that answered; an alias request returns a versioned ID such as jev-1.14.0 |
answers | object | Answers keyed by the question names in the request |
answers.*.type | string | Question type: choice, score, noul, or box |
answers.*.choice | string | Choice: option with the highest probability |
answers.*.score | number | Score: probability-weighted expected level; may fall between integer levels, e.g. 1.87 |
answers.*.noul | number | Probability of yes, 0โ1. Near 1 favors yes; near 0 favors no; near 0.5 indicates uncertainty. |
answers.*.box | object | top, bottom, left, and right edges as fractions of image height or width, 0โ1, measured from the top-left corner |
answers.*.confidence | number | Confidence for Choice, Score, and Box, 0โ1; higher means more certain. Box confidence may be null. |
answers.*.probabilities | object | Choice: probability per option. Score: probability per level, keyed by its index. Totals approximately 1. |
answers.*.legend | object | Score: mapping of level indices to descriptions |
usage.input_tokens | integer | Input token count; basis for billing |
usage.output_tokens | integer | Output token count; currently not billed |
Using confidence: Probabilities and confidence are calibrated and can support automated decision thresholds. For example, automatically process
noul โฅ 0.8, send 0.2 ~ 0.8 for human review, and allow < 0.2. Tune thresholds on a set of manually labeled examples.Billing
Formula: Cost per request =usage.input_tokens / 1,000,000 ร input price
| Model | Input price (USD / million tokens) | Output |
|---|---|---|
jev-1.14.0 | 0.042 | Free |
jev-1.13.0 | 0.042 | Free |
jev-latest | 0.042 | Free |
jev-preview | 0.042 | Free |
- These are official base prices. Actual charges also apply the multiplier for the API keyโs group. Refer to the prices shown in the model marketplace.
- Input tokens include
state, all question instructions, options, and level descriptions, plus image tokens when images are included. output_tokensis informational and currently not billed.- Long context keeps the same per-token price and only permits longer inputs.
input_tokens: 443, the cost is 443 / 1,000,000 ร 0.042 โ USD 0.0000186, or about USD 0.19 for 10,000 such requests.
Error handling
Failed requests return a JSON error object. The platform may wrap upstream errors:message may contain a specific reason or only openai_error. Do not rely on this field alone to identify the cause. Example error format:
{
"error": {
"message": "max_tokens_exceeded",
"type": "bad_response_status_code",
"param": "",
"code": "bad_response_status_code"
}
}
Official API validation failures may return 422, while a gateway may return 400 with an
error wrapper. Check the HTTP status first, then read the error fields. Do not assume every gateway preserves the official detail structure or specific error messages.| Issue | Suggested action |
|---|---|
| Invalid API key | Check Authorization: Bearer ... |
Missing model | Include model in the JSON body or the multipart request part |
Missing state / questions | Both are required; questions must contain at least one question |
| Choice error | Supply criteria as an object with 1โ255 options |
| Score error | Supply criteria as an array with 1โ10 levels |
| Noul error | Supply at least one of instructions or criteria |
| Unknown question type | type must be choice, score, noul, or box |
| Image error | Send TypeSafe-Danger-Zone-Images: true; both dimensions must be 280โ1568 pixels. Use still PNG / JPEG and Base64 without a data:image/...;base64, prefix. See Image limits. |
| Box error | Use jev-1.14.0 or an alias currently pointing to it, with an image and the Images header |
openai_error | No specific cause was exposed. Check image dimensions, model/type compatibility, required fields, and context length. If needed, contact the platform with the request time and request ID. |
max_tokens_exceeded | Reduce state or send TypeSafe-Danger-Zone-Long-Context: true |
/v1/chat/completions fails | Jev is not a chat model; use POST /v1/systemone |
