Quickstart

Cobble is an OpenAI-compatible inference API. If your code or tools speak to OpenAI, they speak to Cobble — change the base URL and key, and you're running on recycled silicon and renewable energy.

You'll make your first request in about two minutes.

1. Get your API key

Create a key in the dashboard under API Keys. Store it as an environment variable — never hard-code it:

export COBBLE_API_KEY="sk-cobble-..."

2. Point at the API

Everything lives under one base URL:

https://api.cobble.network/v1

Standard OpenAI-compatible routes: /v1/chat/completions, /v1/models, /v1/embeddings.

3. Make your first request

curl https://api.cobble.network/v1/chat/completions \
  -H "Authorization: Bearer $COBBLE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen/qwen3.6-27b",
    "messages": [{"role": "user", "content": "Say hello from Cobble."}]
  }'

If you get a JSON response with a choices array, you're live. If not, see Troubleshooting below.

Use an SDK

Any OpenAI SDK works — just override the base URL.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.cobble.network/v1",
    api_key=os.environ["COBBLE_API_KEY"],
)

response = client.chat.completions.create(
    model="qwen/qwen3.6-27b",
    messages=[{"role": "user", "content": "Say hello from Cobble."}],
)
print(response.choices[0].message.content)

JavaScript / TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.cobble.network/v1",
  apiKey: process.env.COBBLE_API_KEY,
});

const response = await client.chat.completions.create({
  model: "qwen/qwen3.6-27b",
  messages: [{ role: "user", content: "Say hello from Cobble." }],
});
console.log(response.choices[0].message.content);

Streaming, tool calling, and JSON mode all work through the standard OpenAI request parameters (stream: true, tools, response_format).

Connect a coding agent

Cobble works with the major open coding agents out of the box:

• opencode — terminal coding agent, our recommended starting point
• Hermes Agent — auto-discovers the full Cobble catalog
• OpenClaw — personal AI assistant
• Cursor — chat panel integration (read the caveat first)

How limits work

Your plan covers chat requests inside a rolling 5-hour window — unmetered, no per-token counting. Requests beyond the window, plus OCR and embeddings, draw from your wallet at the published API rates. You control a spend cap in the dashboard; with no wallet spending enabled, requests past the window return 429 until the window rolls forward.

Troubleshooting

Still stuck? Contact us — include the request ID from the error response.