Replace your LLM provider

Swap your existing OpenAI-compatible provider for Corti Models in two changes.

Already using the OpenAI SDK, Azure OpenAI, or any OpenAI-compatible provider? This guide shows you how to point that existing client at Corti Models, then make your first request. No new SDK, rewrite, or migration.

This is the direct API path for replacing OpenAI or Anthropic in your application. If you want to use Corti Models as the backend for your coding agent (OpenCode, Pi, Crush, or ForgeCode), see Use Corti Models in your coding agent instead.

US projects: corti-s1 model IDs are served by Azure-hosted OpenAI models. See Regions.

Get your API key

Corti Models accepts three credential types as the Bearer token:

• Project API key — generated from an API Client in the Corti Console. Used in this quickstart.
• User API key — your personal key from the Corti Models page in the Console.
• OAuth 2.0 access token — short-lived tokens from the standard Corti client credentials flow. See Authentication.

Grab a project API key from the Corti Console:

Open API Clients in the Corti Console

Sign in, pick a client, then Copy as → Corti Models Service API Key.

Swap your provider

If your code looks like this today:

Python

import os

from openai import OpenAI

client = OpenAI(
    api_key=os.environ["OPENAI_API_KEY"],
)

response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}],
)

Change the base URL and API key, and you’re running on Corti:

Python

from openai import OpenAI

# Replace these with your values
API_KEY = "<your-api-key>"
# corti-s1 | corti-s1-instant | corti-s1-mini | corti-s1-mini-instant
MODEL = "corti-s1"
REGION = "<eu-or-us>"

client = OpenAI(
    base_url=f"https://ai.{REGION}.corti.app/v1",
    api_key=API_KEY,
)

response = client.chat.completions.create(
    model=MODEL,
    messages=[{"role": "user", "content": "Hello!"}],
)

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

Everything else in your code — streaming, tool calling, JSON mode, multi-turn conversations, the Responses API — works unchanged. Corti Models is a drop-in replacement.

corti-s1-mini and corti-s1-mini-instant also accept image inputs for OCR, image descriptions, and UI analysis. See Image input on the Models page, or the example below.

Send an image

corti-s1-mini and corti-s1-mini-instant accept image inputs via the standard OpenAI multimodal content format. Encode your image as a base64 data URI and pass it in the messages array:

JavaScript

import OpenAI from "openai";
import { readFileSync } from "fs";

// corti-s1-mini | corti-s1-mini-instant
const MODEL = "corti-s1-mini";

// Replace these with your values
const API_KEY = "<your-api-key>";
const IMAGE_PATH = "<path-to-your-image>";
const REGION = "<eu-or-us>";

const b64 = readFileSync(IMAGE_PATH).toString("base64");

const client = new OpenAI({
  baseURL: `https://ai.${REGION}.corti.app/v1`,
  apiKey: API_KEY,
});

const response = await client.chat.completions.create({
  model: MODEL,
  messages: [{
    role: "user",
    content: [
      { type: "text", text: "Describe this image." },
      { type: "image_url", image_url: { url: `data:image/jpeg;base64,${b64}` } },
    ],
  }],
});

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

See which models are available

List the models your credentials can access to confirm the model IDs you can pass:

cURL

# Replace these with your values
API_KEY="<your-api-key>"
REGION="<eu-or-us>"

curl "https://ai.${REGION}.corti.app/v1/models" \
  -H "Authorization: Bearer ${API_KEY}"

See the Models page for the full lineup and pricing.

Next steps

Models

Compare the four model variants by capability, reasoning, speed, and price.

Use Corti Models in your coding agent

Connect OpenCode, ForgeCode, Crush, or Pi to Corti Models with the Corti CLI.

API Reference

Browse the full Corti Models API with interactive examples.

Authentication

How authentication works at Corti.