# OpenAI Responses with Keln
Use the OpenAI Responses API with Keln to run open models in agents and other applications. Set
Keln's base URL and API key, then choose a model. The API uses the same models and pricing as Chat
Completions.
To use Keln with the Codex app or CLI, see [Codex with Keln](codex.md).
## What you need to start
- A Keln API key. Sign up at [app.keln.ai](https://app.keln.ai), then open **Keys** →
**Create key**.
- Credit on your Keln account. **Billing** → **Add credit**.
- Your key saved in the `KELN_API_KEY` environment variable.
## Make a request
```bash
curl https://api.keln.ai/v1/responses \
-H "Authorization: Bearer $KELN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "moonshotai/kimi-k3",
"input": "Say hello from Keln.",
"stream": true
}'
```
## Keep the conversation in your client
Keln serves the stateless Responses subset. Your client keeps the conversation, and each request
carries the full `input`. Keln returns a named `400` error for `store: true`,
`previous_response_id`, and an `input` containing an `item_reference`.
## Choosing a model
Use a Keln model id from [Models & pricing](models-pricing.md).
## API reference
The [Responses API](api/responses.md) page lists the parameters, the response object, and the
streaming events.
OpenAI Responses with Keln
Use the OpenAI Responses API with Keln to run open models in agents and other applications. Set
Keln's base URL and API key, then choose a model. The API uses the same models and pricing as Chat
Completions.
Keln serves the stateless Responses subset. Your client keeps the conversation, and each request
carries the full input. Keln returns a named 400 error for store: true,
previous_response_id, and an input containing an item_reference.