Documentation

Build with Cinder

Cinder is a fast, private AI chat product and an OpenAI-compatible API, powered by deepseek-v4-flash over OpenRouter. These docs cover getting an API key, making your first request, and the full chat-completions reference.

Preview. The Cinder developer API is in preview. Endpoints are OpenAI-compatible, so existing OpenAI SDKs work by changing the base URL and key. Manage keys on the developer platform.
🚀

Quickstart

From zero to your first streamed response in under five minutes.
⌗

API reference

The /v1/chat/completions endpoint, every parameter, and response shapes.
🧠

Models & pricing

Context window, capabilities, and token pricing for the model lineup.
🔑

Developer platform

Create API keys, watch usage, and top up your balance.

What is Cinder?

Cinder began as a clean LLM chat app — passwordless accounts, synced history, collapsible reasoning, a live context and cost meter — and exposes the same model through a developer API so you can build it into your own apps.

  • OpenAI-compatible. Point any OpenAI SDK at the Cinder base URL and your key.
  • Streaming first. Server-sent events stream tokens (and reasoning) as they arrive.
  • One model, one key. No model zoo to manage — deepseek-v4-flash with a 1M-token context window.

Base URL

All API requests go to:

https://cinder-pn3.pages.dev/v1

Authenticate with a bearer token — your API key from the platform: 

Authorization: Bearer sk-cinder-...

Next steps

Make your first request →

Install an SDK and stream a completion.

Read the API reference →

Parameters, streaming, and error codes.