Quickstart

In just a few minutes, you’ll be chatting with the latest open source models and sending inference requests with the Modular Cloud API.

Try the playground

Here’s how to play with some models without writing any code.

Create an API key

Click API keys in the sidebar and create a new Personal key.Make sure to enable Model access, and copy the key to a safe place—you’ll need it again below to call the API.

Chat with a model

Click Playground in the sidebar and select your API key at the top.Now start chatting with a model!

You can try different models using the menu at the top of the chat.

Run inference with the API

Here’s a quick guide to running inference with the chat completions API.

Set your API key

Insert the API key you created above or create a new key.

export MODULAR_API_KEY="your_api_key"

$env:MODULAR_API_KEY="your_api_key"

Install the OpenAI client

You can skip this if you’re going to use cURL.

pip install openai

uv init quickstart && cd quickstart
uv add openai

pixi init quickstart && cd quickstart
pixi add openai

mkdir quickstart && cd quickstart
npm init -y && npm install openai

Copy this code

import os
from openai import OpenAI

client = OpenAI(
  base_url="https://api.modular.com/v1",
  api_key=os.environ.get("MODULAR_API_KEY"),
)

response = client.chat.completions.create(
  model="google/gemma-4-31b-it",
  messages=[{"role": "user", "content": "Hello, how are you?"}],
)

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

import process from "node:process";
import OpenAI from "openai";

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

async function main() {
  const response = await client.chat.completions.create({
    model: "google/gemma-4-31b-it",
    messages: [{ role: "user", content: "Hello, how are you?" }],
  });

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

main();

curl -X POST https://api.modular.com/v1/chat/completions \
  -H "Authorization: Bearer $MODULAR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "google/gemma-4-31b-it",
    "messages": [{"role": "user", "content": "Hello, how are you?"}]
  }'

Run it

python main.py

uv run main.py

pixi run python main.py

npx tsx main.ts

You should quickly see the response.

That’s it! You’re up and running. To try a different model, just change the code’s model parameter: Browse the available shared endpoints and click a model, then copy the Model ID shown on the right.

Next steps

Check out these guides to learn more about using the Modular Cloud APIs:

Text inference: text-to-text, image-to-text, and video-to-text
Image inference: generate and transform images with FLUX
Video inference: generate video from text with Wan

Get started

Inference

Administration

Try the playground

Run inference with the API

Next steps

​Try the playground

​Run inference with the API

​Next steps

Try the playground

Run inference with the API

Next steps