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

# Understand token usage

> Learn what basic and premium tokens are, what actions consume them, and how to read usage.

daoco uses two token types to meter usage: **basic tokens** and **premium tokens**.

## Basic tokens

Basic tokens power chat and text-heavy work.

* New threads start with a base chat charge.
* Longer conversations can use more basic tokens as context grows.
* Content drafts, brand summaries, post generation, SEO help, campaign planning, and research can use basic tokens.
* If you stop a run after daoco has already processed context or generated output, the consumed basic tokens can still count. Stop prevents more work; it does not undo work already processed.

## Premium tokens

Premium tokens cover heavier media and generation work, such as image generation, logo exploration, video clipping, and rendering-heavy workflows.

## Context meter in chat

Every chat thread has a **context meter** in the composer, next to the send button. It shows how full the active context window is and, once the thread has billed usage, how many **basic tokens** the thread has charged so far.

Hover or focus the meter to open the usage tooltip.

### Circle

* The ring fills as the thread uses more of its **current context** window.
* When billing has started, the number in the center is how many **basic tokens** this thread has charged so far. Large values are shortened (for example `1.2k`).
* The ring may turn **amber** when context is nearly full, compaction is running, or the assistant is paused at a safe boundary. It may turn **green** briefly after context is successfully refreshed.

### Tooltip

The tooltip has two layers of information:

**Current context** — what daoco can actively work with on the next turn.

* **Percentage** — how full the working window is.
* **Progress bar** — the same percentage, shown visually.
* **`15,808 / 375,000 tokens`** — tokens currently in context vs the thread's context limit.

Below that, a single line summarizes **cumulative usage for the whole thread**:

* **`in:`** — input tokens (your messages, retained history, attachments, and tool context sent to the model).
* **`out:`** — output tokens (assistant responses in this thread).
* **`total:`** — combined input and output for the thread. This value is emphasized in the tooltip.

Current context and thread totals measure different things. After compaction, current context may drop while **total** still reflects everything daoco has processed in the thread.

### Compaction

When the thread gets long, daoco can compact older messages into a summary so the conversation can continue. You may briefly see compaction status in the tooltip. See [Context management](/tips/context-management).

## Queued follow-ups and stopped runs

Follow-up messages queued while a run is active do not start a separate run immediately. They begin using tokens when daoco starts processing them.

Stopped, timed-out, or failed runs may still show usage if daoco already processed part of the request. This keeps the context meter and billing history aligned with the work that actually ran.

## Review usage

Open **Settings** -> **Billing** to review monthly token quotas, plan state, and usage. Growth workspaces can use billing analytics to understand usage across brands and users when the workspace role has access.

## Reduce avoidable usage

* Start focused threads for distinct projects.
* Provide clear constraints in the first prompt.
* Reuse existing assets and brand context.
* Ask for one revision at a time when narrowing output.