# Model inference

Select a model and generate replies, embeddings, or reranking results. Use the corresponding management task to store generation records or maintain local conversations.

## Choose a task

| Task | Page | Description |
| --- | --- | --- |
| Find and select a model | [Find and select a model](models.md) | Check models, capabilities, and prices available to the current credential. |
| Generate replies with Chat Completions | [Generate replies with Chat Completions](chat-completions.md) | Send chat messages and read complete or streamed replies. |
| Generate replies with Messages | [Generate replies with Messages](messages.md) | Optionally count input tokens and generate message replies. |
| Generate and manage results with Responses | [Generate and manage results with Responses](responses.md) | Generate content and, when supported, store, retrieve, and delete results. |
| Manage conversations and items | [Manage conversations and items](conversations.md) | Maintain local conversation metadata and item summaries. |
| Generate embeddings | [Generate embeddings](embeddings.md) | Convert single or batch text inputs into vectors. |
| Rerank documents | [Rerank documents](rerank.md) | Rerank existing candidates by relevance to a query. |

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

Find and select a model <models>
Generate replies with Chat Completions <chat-completions>
Generate replies with Messages <messages>
Generate and manage results with Responses <responses>
Manage conversations and items <conversations>
Generate embeddings <embeddings>
Rerank documents <rerank>
```