Change the base URL. Keep your client.

Quickstart

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.COVIBE_KEY,
  baseURL: "https://covibe.me/v1"
});

await client.chat.completions.create({
  model: "coding-prod",
  messages: [{ role: "user", content: "Fix this test" }]
});

Providers

OpenAI, Gemini and Grok use the same Covibe base URL. Add upstream accounts in the dashboard, place them in a provider group and assign that group to a downstream key.

{
  "base_url": "https://covibe.me/v1",
  "api_key": "$COVIBE_KEY",
  "providers": ["openai", "gemini", "grok"]
}

Call /v1/models with the downstream key and use a returned model ID. Grok access depends on the connected xAI accounts and their available models.

Model aliases

{
  "coding-prod": "budget rule + fallback",
  "coding-fast": "fast default",
  "demo-safe": "demo cap, no premium"
}

Policy result

{
  "model": "coding-prod",
  "policy": "budget rule",
  "budget": "team-agents",
  "fallbacks_available": 2,
  "quota_remaining": "68%"
}

Deploy

Run the gateway behind HTTPS, keep raw provider ports private and give apps only the Covibe base URL. Grok is served through the main /v1 gateway; no separate public Grok port is required.