Docs Get started in minutes

Quickstart guide

Create a key, set the base URL and make your first request, in a few minutes.

Connecting to DaLAi takes three things: create a key, set the base URL in your tool, and choose a model name. You do not need to install any library.

1. Create a key

Keep your key as you would a password. Anyone who has it spends from your balance. If it leaks anywhere, the same page lets you delete it and create a new one.

2. Base URL

In your tool's settings, set the base URL to this value:

https://dalai.ir/v1

Any tool that works with the standard OpenAI API will work with no other change. The endpoints are described in full in API overview.

A complete example with curl:

curl https://dalai.ir/v1/chat/completions \
  -H "Authorization: Bearer sk-YOUR-KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "claude-haiku-4-5", "messages": [{"role": "user", "content": "Hello"}]}'

3. Set up your tools

Change only these fields. All other settings stay at their default values.

Tool	What to set
Cline	Set API Provider to OpenAI Compatible, set Base URL to `https://dalai.ir/v1`, then enter your key
Continue	In `config.yaml`: `provider: openai` together with `apiBase: https://dalai.ir/v1`
Aider	`aider --openai-api-base https://dalai.ir/v1 --openai-api-key sk-...`
Open WebUI	Settings → Connections → OpenAI API, and the URL `https://dalai.ir/v1`
LibreChat	A custom endpoint in `librechat.yaml` with `baseURL: https://dalai.ir/v1`
Jan	Settings → Model Providers, and add an OpenAI-compatible provider

A fuller list, and what has been tested, is in Tools overview.

Choosing a model

Models are offered under public names, such as claude-opus-4-6 or qwen3-coder-480b. If the service behind a name changes, the name stays the same and you change nothing.

To choose between them, see Models overview; for live prices, see Model Square.

Usage and cost

Your balance is prepaid and shown in toman. The cost of each request is based on the tokens it actually used, so a short question costs a fraction of what a long conversation does.

Every request is recorded individually in the Console, under Usage Logs: which model answered, how many input and output tokens it used, and how much it cost. Full details are in Wallet, top-ups and billing.

If something does not work

Check the model name against the model list; a wrong name is the most common cause.
Do not put your key in public code or a shared repository.
For long responses, streaming is supported, and it is the default in most tools.

The meaning of each error message is explained in Errors and limits.