# Integrate with apps

Integration moves a model from console experimentation into your application. You need three values: a model ID, a Genesis API Key, and the Base URL shown on the **Usage Guide** page. Genesis API requests authenticate with `Authorization: Bearer <Genesis API Key>`; do not use an AI Studio personal access token, a UC Control API Key, or a browser cookie as the model API credential.

## Copy connection details and code samples

1. Open **Usage Guide** in the left sidebar.
2. Copy the **Base URL** from **Connection information**.
3. In **API usage**, choose an endpoint and language. The page currently offers cURL, Python, and Node.js samples; available endpoints follow the page dropdown.
4. Copy the generated code and replace the API Key, Base URL, and `model` with your values.
5. Run the request locally and confirm the business result and usage fields returned by the endpoint.

Current samples on the Usage Guide page cover chat, multimodal chat, Responses, Messages, embeddings, rerank, and the model list. For full parameters, response fields, and endpoints outside the code generator, see [Developer · Genesis API](../../developer/api/genesis-model-api/index.md) and [Reference · Model API](../../reference/api/model-api.md).

## Compatible format and authentication

| Item | Value |
| --- | --- |
| Compatible format | OpenAI API. Point the official SDK `base_url` to Genesis, set `api_key` to a Genesis Key, and leave the remaining call code unchanged. |
| Authentication | Only `Authorization: Bearer sk-gsa-v1-...` (or a locally issued Genesis Key with the same header). Rejected alternatives include `X-API-Key`, UC Control API Keys, AI Studio personal access tokens, and browser cookies. |

## What to check first when integration fails

| Symptom | Check first |
| --- | --- |
| 401 or auth failure | Whether the API Key is valid, whether you use `Authorization: Bearer`, and whether you mixed in another product credential. |
| No model access | Whether the Key's allowed model scope includes the requested model ID. |
| 429 or quota/rate errors | Whether the Key's quota, RPM, TPM, or concurrency limits were reached. |
| Unsupported endpoint or parameters | Whether the model supports the endpoint; verify capabilities under **Models**, then retest in **Debug**. |
| Need the exact request | Open **Logs**, filter by time, key, model, or status, and inspect the request details. |