Bring your own models
Typillar generates code with a large language model, but it never runs that model on its own servers. Inference happens on your side — either through Cloudflare Workers AI or through a model provider key you supply. This page explains why, and what you can choose.
Two ways to provide inference
Section titled “Two ways to provide inference”Cloudflare Workers AI
Section titled “Cloudflare Workers AI”If you connect Cloudflare, you can run inference on Workers AI — models hosted on Cloudflare’s network, billed through your Cloudflare account. There’s nothing extra to configure beyond the Cloudflare connection itself.
Your own provider key
Section titled “Your own provider key”Prefer a specific model or vendor? Supply your own API key and Typillar will use that provider for inference. Your key is stored encrypted and used only to run your builds. See Models & API keys for how keys are handled.
Why inference runs on your side
Section titled “Why inference runs on your side”- Cost transparency. You see and pay for inference directly, on your own account — no markup, no proxy, no surprise line item.
- Control. You choose the model and provider, and you can change or revoke the key at any time.
- A clean trust boundary. Because Typillar never runs the model, your prompts and generated code don’t pass through our inference servers. See Zero inference on our servers.
Choosing between them
Section titled “Choosing between them”| If you want… | Use |
|---|---|
| The simplest setup | Cloudflare Workers AI |
| A specific model or vendor | Your own provider key |
| Everything billed in one place | Cloudflare Workers AI |
| To reuse an existing model contract | Your own provider key |
You can start with Workers AI and switch to your own key later — it doesn’t change how the loop works, only where the tokens are generated.