Get to first token in five minutes.

Quickstart

Every request goes to https://api.idclinks.com/v1. Set your API key in the Authorization header and pick a model from the catalog.

curl https://api.idclinks.com/v1/chat/completions \
  -H "Authorization: Bearer $IDCLINKS_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-4.6",
    "messages": [
      {"role": "user", "content": "Hello, world."}
    ]
  }'

Authentication

Bearer-token authentication. API keys are scoped to a project; project-scoped keys can be rotated independently and audited from the dashboard.

Model names

Pass the model as a string. Names are stable across regions; you don't need to encode the region or replica in the model name.

Chat completions

POST /v1/chat/completions. Request and response shapes mirror the OpenAI schema. The full message array, role types, and parameter list are unchanged.

Streaming

Set stream: true. Responses arrive as text/event-stream with standard SSE delta chunks. The final chunk is [DONE] exactly as in the upstream schema.

Tool calls

Pass a tools array on the request. Models that support tool-use will emit tool_calls on the assistant message; reply with a tool-role message and the result. Available on Claude, GPT, Gemini, Mistral, and Command R+.

Structured output

Set response_format: { type: "json_schema", json_schema: { ... } }. The gateway validates against your schema and retries once on a validation miss before surfacing the error.

Vision input

Pass image content as image_url parts inside the message. Public URLs and base64 data URIs are both supported. Vision-capable models are flagged in the catalog with the vision tag.

Response headers

Every response includes routing metadata so you always know where your tokens were generated and how much they cost.

Errors & retries

The gateway returns standard HTTP status codes. 429 and 5xx responses include a Retry-After header; client-side retries should respect it. Upstream failovers are automatic and don't count against your retry budget.

• 400 — malformed request (validation failed before upstream)
• 401 — invalid or missing API key
• 402 — payment required (trial credit exhausted)
• 429 — rate-limited, retry after the header value
• 503 — all upstream pools exhausted; rare

Rate limits

Default Developer-tier ceiling is 10,000 requests/min per project. Scale and Enterprise plans get reservable per-tenant budgets that you can rebalance from the dashboard or via the management API.

Observability

Per-request traces are emitted to your sink of choice:

• Webhook — POST to your endpoint, signed with HMAC-SHA256.
• S3 — newline-delimited JSON, written to a prefix you own.
• OpenTelemetry — OTLP/gRPC export to a collector you specify.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.idclinks.com/v1",
    api_key=os.environ["IDCLINKS_KEY"],
)

resp = client.chat.completions.create(
    model="claude-sonnet-4.6",
    messages=[{"role": "user", "content": "Hi"}],
)
print(resp.choices[0].message.content)

TypeScript / Node

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.idclinks.com/v1",
  apiKey: process.env.IDCLINKS_KEY,
});

const resp = await client.chat.completions.create({
  model: "gpt-5.5",
  messages: [{ role: "user", content: "Hi" }],
});

cURL

curl -N https://api.idclinks.com/v1/chat/completions \
  -H "Authorization: Bearer $IDCLINKS_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4",
    "messages": [{"role":"user","content":"Hi"}],
    "stream": true
  }'