Usage Explorer¶
The Usage Explorer provides users visibility into graphics processing unit (GPU), central processing unit (CPU), and large language model (LLM) API usage across the platform, providing general usage information broken down by service.
The following services are tracked in the Usage Explorer:
- Modeling
- Inference
- NVIDIA AI Enterprise
- Vector Database Creation
- GenAI Playground
- Custom Models
- Moderations
- Data Management
- Predictions
Note
The services displayed in the Usage Explorer may vary depending on the type of usage being viewed.
To access the Usage Explorer, open Account settings > Usage Explorer. From here, you can view resource consumption by service for a given date range, as well as export the report as a .csv file.
| Element | Description | |
|---|---|---|
| 1 | Date range selector | Use the two fields to display usage information for a specific date range. |
| 2 | Export | Download the report as a .csv file. |
| 3 | Usage options | Select the usage information to view—GPU, LLM, or CPU. |
GPU usage¶
The GPU Usage page reports data on GPU usage, providing general usage information broken down by service. GenAI features, for example, rely on GPU hardware for a range of workloads related to training, hosting, and running inference on LLMs.
To access the GPU Usage page, click GPU Usage in the Usage Explorer.
This page consists of a table that displays the name of the service using the resources, as well as the following details about each task under that service:
| Field | Summary |
|---|---|
| Service | The service that used the GPU (e.g., Notebooks, Modeling, Vector Database Creation, etc.). |
| Resource name | The number of GPUs, as well as the number of CPUs and amount of RAM/VRAM, used by the task. |
| Cloud | The cloud provider executing the resource. |
| Region | Region where costs are computed. |
| Unit cost (per 1 hour) | The total cost of the resource each hour. |
| Usage quantity | The amount of time each task used the resources within the specified time period. |
| Amount | The current cost of the task. |
LLM API usage¶
The LLM API Usage page reports data on which LLM models are being used by which services, as well as how much each model is being used. The page provides general usage information broken down by service, tracking token consumption for each task. This detailed monitoring helps identify which services are consuming the most LLM resources across various services.
To access the LLM API Usage page, click LLM API Usage in the Usage Explorer.
This page consists of a table that displays the name of the service using the resources, as well as the following details about each task under that service:
| Field | Summary |
|---|---|
| Service | The service that used the LLM API (e.g., Custom Models, GenAI Playground, etc.). |
| Resource name | The name of the LLM model called via the API. |
| Cloud | The cloud provider executing the resource. |
| Region | Region where costs are computed. |
| Unit cost (per 1 hour) | The total cost of the resource each hour. |
| Usage quantity | The number of tokens used by the task. |
| Amount | The current cost of the task. |
CPU usage¶
The CPU Usage page provides an overview of central processing unit (CPU) broken down by service. This allows for easy monitoring of exactly which users and services are consuming the most CPU resources, potentially helping to identify areas for optimization or budgetary concerns.
To access the CPU Usage page, click CPU Usage in the Usage Explorer.
This page consists of a table that displays the name of the service using the resources, as well as the following details about each task under that service:
| Field | Summary |
|---|---|
| Service | The service that used the CPU (e.g., Custom Inference Model, Data Management, etc.) |
| Resource name | The number of processors and amount of RAM used by the task. |
| Cloud | The cloud provider executing the resource. |
| Region | Region where costs are computed. |
| Unit cost (per 1 hour) | The total cost of the resource each hour. |
| Usage quantity | The amount of time each task used the CPU. |
| Amount | The current cost of the task. |