> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orq.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Billing and usage tracking
> Monitor LLM usage, costs, and billing cycles across workspaces. Track token consumption, model generation metrics, and spending analytics for AI applications.
## Workspace Usage
Billing can be accessed inside the **Organization** panel, where you have an overview of your usage over the current billing cycle.
A graph displays the number of **LLM**, **Retrievals**, and **Cache** over time in your workspace. At the top-right of the graph, you can see your current usage against your plan capacity. When going beyond capacity, **additional events are added** to your billing cycle.
View detailed usage breakdowns, trends, and event analytics in the **Analytics** dashboard.
## Seats
On the **Growth** plan, the number of members you can invite is controlled by your seat count. Seats are managed directly from the **Organization → Billing** page.
### Adding and Removing Seats
Use the **Add seat** button or the icon on the billing page to add or remove seats. Each seat corresponds to one workspace member slot. Seat changes are reflected immediately and will update your billing accordingly.
You must have an available seat before inviting a new member. To learn how to invite members, see [Members and Teams](/docs/administer/permissions/overview).
## Understanding Trace Storage Usage
Our platform stores distributed [**traces**](/docs/observability/traces) received through OpenTelemetry:
1. Each trace is composed of one or more **spans**, which represent individual operations or segments of a request.
2. Each span is encoded as **JSON data** before being stored.
3. Once encoded, we apply an **indexing layer** on top of the raw data. This indexing allows for fast search and filtering across large trace datasets.
To account for both the raw and indexed data, we estimate total storage consumption as:
```
total_storage = (raw_bytes * 1.7)
```
In other words, the indexed representation roughly doubles the storage footprint compared to the raw JSON data.
From our analytics:
* **Total traces:** 100
* **Total storage consumed:** \~10 MB
* **Average size per trace:** \~120 KB
This average reflects the JSON encoding plus the indexing overhead.
Using OpenTelemetry, some exporters send **incremental payloads** which might result in **sudden increases**.
* This means each new span **can contain data from previous messages** as well.
* This results in **larger payloads over time**, and consequently, sudden jumps in reported storage usage.
* Check your exporter configuration when dealing with sudden increases.
To **reduce storage usage**, consider:
* Sampling fewer spans per trace.
* Filtering out high-volume, low-value telemetry data.
* Using compression or limiting payload size before sending.
These optimizations can help maintain observability while keeping your storage footprint efficient.
## Understanding Events
A single [Deployment](/docs/deployments/overview) invoke contains multiple events, each event will incur costs reflected in your Billing and Plan Usage.
To understand better the events held within your Deployments, lookup [Analytics](/docs/analytics/dashboards) and explore the events embedded into each generation.
## Rate Limits
Our APIs are protected through Rate Limits on a per-account basis to ensure **fair and efficient use of the API**. This helps maintain optimal performance and prevent server overload, while also protecting against potential abuse and limiting costs effectively.
When reaching rate limit, API calls are denied with a **429 Too Many Requests** response.
| Subscription | Rate Limit |
| ------------ | ----------------------- |
| Developer | 50 requests per day |
| Growth | 100 requests per minute |
| Enterprise | Custom |
To learn more about the Orq.ai Pricing options or to upgrade your plan, see [Our Pricing Page](https://orq.ai/pricing).