Skip to main content

Call a model API

Add hosted AI models to your application with an OpenAI-compatible API.

Run code in a Sandbox

Give your agent an isolated environment to create a sandbox and run code.

Choose how you want to call a model

Send your first request

Create an API key, send a small LLM request, and verify the response yourself.

Use your agent

Ask an agent to help connect to Novita AI, then verify the real API response.

Before you run

No install — paste this into your agent chat:
Optional — install once if your agent supports skills:
The novita-ai skill covers model APIs, GPUs, Agent Sandbox, integrations, and troubleshooting. Your agent still needs a valid API key and available credit when it makes a request.View the skill on GitHub →

Explore Novita AI

Model APIs

Explore hosted models and API capabilities.

GPUs

Run and manage your own models on dedicated GPUs.

API Reference

Look up endpoints, parameters, authentication, and responses.

API errors

Find causes and fixes for common API errors.

API key management

Create, store, and configure API keys.

Billing

Check credit, usage, and payment settings.
Last modified on October 10, 2026