# Get started

Use this section to complete your first Genesis model call. Prepare a usable credential and model ID, send a minimal Chat Completions request, and inspect the response. Then choose an OpenAI-compatible, Anthropic-compatible, or retrieval and vector interface for the application task.

## Genesis tasks

Genesis supports text and multimodal chat, embeddings, and reranking. Chat Completions, Responses, and Messages use different request and response shapes, so use the interface that matches the existing application or client library.

## Recommended order

1. Use [Choose models and capabilities](choose-model-capabilities.md) to select a model type and call format.
2. Call `GET /models` in [Available models](available-models.md) to confirm a model ID available to the current credential.
3. Configure the request header described in [Authentication](authentication.md).
4. Follow the [Chat Completions quickstart](quickstart.md) and read the reply from `choices[0].message.content`.

When the response contains `choices[0].message.content`, the first model reply has been returned. Retain the actual model ID and request ID for later usage, log, or error investigation.

```{toctree}
:hidden:
:maxdepth: 1

Authentication <authentication>
Available models <available-models>
Choose models and capabilities <choose-model-capabilities>
Chat Completions quickstart <quickstart>
```