> ## Documentation Index
> Fetch the complete documentation index at: https://xura.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Models

> Which Claude models Xura supports and when to pick each.

## Supported models

| Model          | ID                  | Best for                                        |
| -------------- | ------------------- | ----------------------------------------------- |
| **Sonnet 4.6** | `claude-sonnet-4-6` | Default. Structured JSON, tool use, most flows. |
| **Opus 4.8**   | `claude-opus-4-8`   | Hardest reasoning, research quality. \~5× cost. |
| **Haiku 4.5**  | `claude-haiku-4-5`  | Fastest, cheapest. Lightweight tasks.           |

## Cost comparison

Rough per-1M tokens (input / output):

| Model      | Input | Output |
| ---------- | ----- | ------ |
| Sonnet 4.6 | \$3   | \$15   |
| Opus 4.8   | \$15  | \$75   |
| Haiku 4.5  | \$1   | \$5    |

## Where each is used by default

* **Flows** — Sonnet.
* **Signals** — Sonnet (was Opus, flipped for cost).
* **Dashboards** — Sonnet.
* **Deep research** — Sonnet.
* **Slides** — Sonnet.

You can override globally in **AI Studio → LLM Configuration**.

## Prompt caching

Enabled on Signals + planner. Anthropic caches the fixed instruction block at \~10% of input cost on hits.