AI Gateway · Quick start

Quick start

Add a provider key, create a NevTan key, and call any hosted model through one OpenAI-compatible endpoint.

The console Quick Start page walks through three steps, with modality tabs across the top — Text, Image, Video, Realtime, and Speech — so you can preview the shape of a call for each kind of model.

The Quick Start page with modality tabs and three numbered step cards — The same three steps in the console, with a modality tab for each kind of model.

1
Add a provider key
Open Bring Your Own Key, pick a provider (e.g. Anthropic), and paste your key. It is encrypted at rest and used only to forward your requests.
2
Create a NevTan API key
Under API Keys, create a key. You call the gateway with this key — never with your raw provider key.
bash
```
 
```
3
Call the gateway
Point the OpenAI SDK at the gateway base URL and choose a provider/model id.
javascript
```
 
```

Or with curl

bash

Tip

Anthropic models take the same OpenAI request shape — the gateway translates to the Messages API and back, so you never touch Anthropic-specific fields.