API protocol
POST · Responses
Direct answerPOST /v1/responses. Authentication, parameters, examples in three languages, result validation and error handling.
Updated · Reviewed
OpenFox · Responses
POST https://api.openfox.vip/v1/responses
Model ID: YOUR_MODEL_ID
Authentication
Authorization: Bearer $OPENFOX_API_KEY
Create a model API key in API Keys. Dashboard sessions and personal access tokens are not model API credentials.
Keep credentials in environment variables on your server. Do not paste real keys, cookies or account records into this guide.
Request parameters
- model (body, string, Required): See schema details.
- input (body, provider-specific, Optional): 输入内容,可以是字符串或消息数组
- instructions (body, string, Optional): See schema details.
- max_output_tokens (body, integer, Optional): See schema details.
- temperature (body, number, Optional): See schema details.
- top_p (body, number, Optional): See schema details.
- stream (body, boolean, Optional): See schema details.
- tools (body, array, Optional): See schema details.
- tool_choice (body, provider-specific, Optional): See schema details.
- reasoning (body, object, Optional): See schema details.
- previous_response_id (body, string, Optional): See schema details.
- truncation (body, string, Optional): See schema details.
Example values are not server defaults. Model-specific parameter support must be confirmed for the selected channel.
Model and channel limitations
Retrieving stored Responses is implemented by task plugins; ordinary relay responses are not universally retrievable.
Request example
Run one language example only. Creation requests may be charged; do not run every example or retry automatically.
Set OPENFOX_REQUEST_ID from the original X-Request-Id when recovering images. Set OPENFOX_TASK_ID from a real task creation response when querying tasks.
curl
# Bash / WSL; curl 7.76+ and jq. Run ONE language example, not all of them.
umask 077
RESULT=$(mktemp)
trap 'rm -f "$RESULT"' EXIT
: "${OPENFOX_API_KEY:?Set OPENFOX_API_KEY}"
curl --silent --show-error --fail-with-body --max-time 180 'https://api.openfox.vip/v1/responses' \
-H "Authorization: Bearer $OPENFOX_API_KEY" \
-H "Content-Type: application/json" \
--data-raw '{"model":"YOUR_MODEL_ID","input":"Hello!","stream":false}' \
--output "$RESULT" || exit 1
jq -e 'type == "object" and (.error | not) and (.success != false) and (.code != false)' "$RESULT" >/dev/null || exit 1
# HTTP/business envelope checked only. Complete result validation is required:
# Require status=completed and non-empty output[].content[] text. A response ID or an in-progress response is not a completed answer.
# Use the Python or JavaScript example in this guide for result processing.
python
# Python 3.10+; install: pip install httpx
import asyncio, json, os
from urllib.parse import quote, urlparse
import httpx
async def main():
url = "https://api.openfox.vip/v1/responses"
headers = {}
headers["Authorization"] = "Bearer " + os.environ["OPENFOX_API_KEY"]
params = {}
async with httpx.AsyncClient(timeout=30, follow_redirects=False) as client:
async def read(target, method='GET', payload=None):
response = await client.request(method, target, headers=headers, params=params, json=payload)
response.raise_for_status()
value = response.json()
if not isinstance(value, dict) or value.get('error') or value.get('success') is False or value.get('code') is False:
raise RuntimeError('API business error')
return value
data = await read(url, "POST", json.loads("{\"model\":\"YOUR_MODEL_ID\",\"input\":\"Hello!\",\"stream\":false}"))
text = ('\n'.join(c.get('text', '') for item in data.get('output', []) for c in item.get('content', []) if c.get('type') == 'output_text') if data.get('status') == 'completed' else '')
if not text.strip():
raise RuntimeError('No completed text output')
print(text)
async def bounded():
await asyncio.wait_for(main(), timeout=180)
asyncio.run(bounded())
javascript
// Node.js 20+; save as example.mjs and run: node example.mjs
const controller = new AbortController();
const timer = setTimeout(() => controller.abort(), 180000);
const signal = controller.signal;
try {
const url = new URL("https://api.openfox.vip/v1/responses");
const headers = {"Content-Type":"application/json"};
if (!process.env.OPENFOX_API_KEY) throw Error('Set OPENFOX_API_KEY');
headers["Authorization"] = "Bearer " + process.env.OPENFOX_API_KEY;
const read = async (target, method = 'GET', payload) => {
const response = await fetch(target, { method, headers, signal, redirect: 'error', body: payload === undefined ? undefined : JSON.stringify(payload) });
if (!response.ok) throw Error('HTTP ' + response.status);
const data = await response.json();
if (!data || typeof data !== 'object' || data.error || data.success === false || data.code === false) throw Error('API business error');
return data;
};
let data = await read(url, "POST", {
"model": "YOUR_MODEL_ID",
"input": "Hello!",
"stream": false
});
const text = data.status === 'completed' && data.output?.flatMap(x => x.content ?? []).filter(x => x.type === 'output_text').map(x => x.text).join('\n');
if (typeof text !== 'string' || !text.trim()) throw Error('No completed text output');
console.log(text);
} finally { clearTimeout(timer); }
Results and success criteria
Require status=completed and non-empty output[].content[] text. A response ID or an in-progress response is not a completed answer.
Error handling
Check both HTTP status and business errors. A timeout does not prove the operation failed. Do not automatically retry creation requests; inspect the original request or usage first.
401/403: check credentials, scopes, model permissions and IP rules. 429: check quota and rate limits. 5xx: preserve a non-secret request ID and contact support.
Python and JavaScript share a 180-second network deadline. Read-only polling is bounded. Unknown or failure states stop the flow. Image decoding is local CPU work with byte and pixel limits.
Uploads, downloads and related endpoints
Only the image editing example uploads a local file. Set OPENFOX_IMAGE_PATH to a supported image. Other examples do not upload files. Download external results without platform credentials. Task artifacts require format-specific verification.
Full request schema
{
"parameters": [],
"requestBody": {
"content": {
"application/json": {
"schema": {
"type": "object",
"required": [
"model"
],
"properties": {
"model": {
"type": "string"
},
"input": {
"description": "输入内容,可以是字符串或消息数组",
"oneOf": [
{
"type": "string"
},
{
"type": "array",
"items": {
"type": "object",
"properties": {}
}
}
]
},
"instructions": {
"type": "string"
},
"max_output_tokens": {
"type": "integer"
},
"temperature": {
"type": "number"
},
"top_p": {
"type": "number"
},
"stream": {
"type": "boolean"
},
"tools": {
"type": "array",
"items": {
"type": "object",
"properties": {}
}
},
"tool_choice": {
"oneOf": [
{
"type": "string"
},
{
"type": "object",
"properties": {}
}
]
},
"reasoning": {
"type": "object",
"properties": {
"effort": {
"type": "string",
"enum": [
"low",
"medium",
"high"
]
},
"summary": {
"type": "string"
}
}
},
"previous_response_id": {
"type": "string"
},
"truncation": {
"type": "string",
"enum": [
"auto",
"disabled"
]
}
}
}
}
},
"required": true
}
}
Sources and verification
docs/openapi/relay.json · controller/plugin_protocol.go · 2026-11-03
SHA-256: aeee6c35a35b2eaced964401f3826f369ad7ac8985c17989dcbbe08337d9fa01
source-reviewed; upstream-not-tested; legacy-schema-needs-field-review
Source review is not upstream business verification. Availability and billing depend on your account, group and channel.
Use cases
- getting-started
- troubleshooting
API protocols
/v1/responses
FAQ
Does this guarantee model availability?
No. Check the exact model, group and endpoint permissions in the marketplace.
Do examples include live credentials?
No. Set environment variables on your server and replace YOUR_MODEL_ID with an available exact model ID.
Should I retry immediately after a timeout?
Do not automatically retry creation requests. Check the request ID, task status and usage to avoid duplicate charges.
Related guides
Official sources
- OpenFox API Official