> For the complete documentation index, see [llms.txt](https://docs.darcyiq.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.darcyiq.com/dispatch/use-dispatch/models.md).

# Models

The model tiers your team can call, what the Models page tells you about each, and how to choose one

In Dispatch you don't call a model by its name. You call a **tier**: a named set of models that DarcyIQ curates and keeps current. You put the tier's id in your request where a model name would normally go. The models behind a tier can be improved over time without you changing a line of code or config.

The **Models** page lists the tiers your organization offers, with what each one costs, how capable it is, and how much it saves against the public model it stands in for. The page is read-only and every role can see it.

{% hint style="info" %}
**Not sure which tier to use?** Start with Auto, marked **picks for you**. Auto routes each request to the cheapest tier that can hold it, choosing between a faster tier and a stronger one step by step.
{% endhint %}

## Using a Tier

Send the tier's id as the `model` in your request. Each row shows the id under the tier's name, and the copy button beside the name copies it.

* **Use the id exactly as shown.** A real model name is rejected, and so is a tier id with a typo.
* **Claude Code sees tiers as `anthropic/<tier id>`.** Claude Code ignores model names without "anthropic" or "claude" in them, so Dispatch lists each tier to it with that prefix. Both spellings reach the same tier.
* **Your tools can discover your tiers.** Dispatch answers the standard model-list request at `/v1/models` with your tiers, so a tool that supports discovery can fill in its model picker by itself.
* **The models behind a tier aren't listed.** Dispatch shows what a tier promises and what it costs, not which models serve it.

The [Cookbooks](/dispatch/use-dispatch/cookbooks.md) page builds every tool's configuration from your tiers, so you rarely need to type an id yourself.

## Reading the Table

| Column                 | What it shows                                                                                                                                   |
| ---------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------- |
| **Tier**               | The tier's name, its id, a short description, and any badge that applies                                                                        |
| **Intelligence index** | An independent capability score out of 100. Auto shows the range between its two tiers.                                                         |
| **Context**            | The largest prompt the tier accepts, in tokens                                                                                                  |
| **Max output**         | The longest reply the tier will produce, in tokens                                                                                              |
| **Input / 1M**         | What you're billed per million input tokens, in USD                                                                                             |
| **Cached input / 1M**  | What you're billed per million input tokens the model has seen recently. When it's the same as the input price, the tier has no cache discount. |
| **Output / 1M**        | What you're billed per million output tokens, in USD                                                                                            |
| **% savings**          | How much less the tier costs than the public model it stands in for, at that model's public price, weighting input and output 3:1               |

A dash means the figure isn't available for that tier, not that it's zero.

Use **Search tiers** to filter by name, id or description. **Sort by** orders the table by **Name** (the default), **Intelligence index**, **Input price**, **Savings** or **As listed**.

### Comparing a Tier with a Public Model

Click a row to open it. Under **Compared with**, Dispatch names the public model the tier is measured against and shows that model's public price for each rate, with how much more or less the tier costs.

### Auto's Prices

Auto's price depends on the work, because each step goes to either its faster tier or its stronger one. Auto's row has a **use case complexity** switch with five settings: **Low**, **Moderate**, **Medium** (the default), **High** and **Frontier**. **Low** prices every step on the faster tier, **Frontier** every step on the stronger, and each setting in between moves a quarter of the steps across. Rates between the two ends are marked with `~` because they're estimates. Hover over a rate to see both ends.

To see what Auto actually chose for your requests, look at **Recent requests** on [Overview](/dispatch/use-dispatch/overview.md).

## Tiers You Can't Call

A tier can be listed but switched off for you. Its row is dimmed and carries a badge saying why:

* **disabled by your organization** — your Dispatch Admin has turned the tier off in [Model Access](/dispatch/manage/manage/model-access.md).
* **disabled by your guardrail** — the guardrail on your keys doesn't allow it. See [Guardrails](/dispatch/manage/manage/guardrails.md).

These tiers stay on the page so you know they exist, but a request that names one is refused. The Cookbooks page leaves them out of its snippets so they never end up in a config file. If you need one, ask your Dispatch Admin.

## Model Intelligence and Price

Two panels below the table help you compare tiers with models you already know.

* **Model Intelligence** ranks each tier by its independent intelligence score from Artificial Analysis, alongside well-known public models, so you can see what each tier sits next to.
* **Intelligence vs. Price** plots the same scores against price per million tokens, blended at three input tokens to every output token. The shaded **Most attractive quadrant** holds what scores above the median and costs below it. This panel appears only when your tiers have prices.

## Choosing a Tier

* **Start with Auto** unless you have a reason not to. Pin a specific tier when you want every request to run on it.
* **Check Context and Max output** for long-running agents or large files. A tier only promises what every model behind it can do.
* **Use Intelligence index and % savings together** to find the cheapest tier that's capable enough for the work.
* **Check what your tool expects.** Some tools need to be told a tier's limits, and the [Cookbooks](/dispatch/use-dispatch/cookbooks.md) page shows each tier's **Limits** for that reason.

## How This Relates to Model Access

This page is the same for everyone. It shows what your organization offers and what you personally can call. A Dispatch Admin decides which tiers the organization may use in **Organization** under **Models**. A change there applies to every key in the organization within a minute. See [Model Access](/dispatch/manage/manage/model-access.md).

## Next Steps

| Goal                                          | Documentation                                                   |
| --------------------------------------------- | --------------------------------------------------------------- |
| Put a tier into your tool's configuration     | [Cookbooks](/dispatch/use-dispatch/cookbooks.md)                |
| Choose which tiers your organization may call | [Model Access](/dispatch/manage/manage/model-access.md)         |
| Restrict tiers for particular people          | [Guardrails](/dispatch/manage/manage/guardrails.md)             |
| See spend and savings by tier                 | [Cost Reporting](/dispatch/cost-and-insights/cost-reporting.md) |


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.darcyiq.com/dispatch/use-dispatch/models.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.