SYC API quickstart

1. Create an API key

Create a private-beta account now. API key creation opens after the production gateway passes authentication, model listing, and completion tests. Keys should only be used from trusted server-side code.

SYC_API_KEY=sk-your-syc-api-key
SYC_API_ENDPOINT=https://api.sycapi.com/v1

Authentication

Send your API key in the Authorization header.

Authorization: Bearer YOUR_SYC_API_KEY

Examples

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_SYC_API_KEY",
    base_url="https://api.sycapi.com/v1"
)

response = client.chat.completions.create(
    model="MODEL_ID_FROM_CONSOLE",
    messages=[{"role": "user", "content": "Compare two model choices for my app"}]
)

print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "YOUR_SYC_API_KEY",
  baseURL: "https://api.sycapi.com/v1",
});

const response = await client.chat.completions.create({
  model: "MODEL_ID_FROM_CONSOLE",
  messages: [{ role: "user", content: "Hello from SYC API" }],
});

console.log(response.choices[0].message.content);

curl https://api.sycapi.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_SYC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MODEL_ID_FROM_CONSOLE",
    "messages": [
      {"role": "user", "content": "Hello from SYC API"}
    ]
  }'

Production retries

Use exponential backoff with jitter for 429, 500, 502, and 503 responses. Set a request timeout, cap retries, and log the request ID when one is returned. Never retry authentication or insufficient-credit errors without fixing the underlying issue.

curl https://api.sycapi.com/v1/models \
  -H "Authorization: Bearer YOUR_SYC_API_KEY"

Supported models

Supported models may change as capacity and provider availability change. Check the customer portal for the current catalog before production use. SYC API is designed as a model adoption layer, not a single-model endpoint.

Error guide

Billing

SYC API uses prepaid credits for API usage. Customers add credits first, reserve a budget when creating a key, and can request refunds for unused credits within the published refund window. This keeps model testing predictable before production traffic grows.