This section explains the concepts of generative AI scale unit (GSU) and
burndown rates. Provisioned Throughput is calculated and priced
using generative AI scale units (GSUs) and burndown rates.

## GSU and burndown rate

A *Generative AI Scale Unit (GSU)* is a measure of throughput for your prompts
and responses. This amount specifies how much throughput to provision a model
with.

A *burndown rate* is a ratio that converts the input and output units (such as
tokens, characters, or images) to input tokens per second, input characters per
second, or input images per second, respectively. This ratio represents the
throughput and is used to produce a standard unit across models.

Different models use different amounts of throughput. For information about the
minimum GSU purchase amount and increments for each model, see [Supported models
and burndown rates](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models) in this document.

This equation demonstrates how throughput is calculated:

    inputs_per_query = inputs_across_modalities_converted_using_burndown_rates
    outputs_per_query = outputs_across_modalities_converted_using_burndown_rates

    throughput_per_second = (inputs_per_query + outputs_per_query) * queries_per_second

The calculated throughput per second determines how many GSUs that you need for
your use case.

## Important Considerations

To help you plan for your Provisioned Throughput needs, review the
following important considerations:

- **Requests are prioritized.**

  Provisioned Throughput customers are prioritized and serviced
  first before on-demand requests.
- **Throughput doesn't accumulate.**

  Unused throughput doesn't accumulate or carry over to the next
  month.
- **Provisioned Throughput is measured in tokens per second, characters per second, or images per second.**

  Provisioned Throughput isn't measured solely based on queries per minute
  (QPM). It's measured based on the query size for your use case, the response
  size, and the QPM.
- **Provisioned Throughput is specific to a project, region, model, and version.**

  Provisioned Throughput is assigned to a specific
  project-region-model-version combination. The same model called from a
  different region won't count against your Provisioned Throughput
  quota and won't be prioritized over on-demand requests.

### Context caching

> [!WARNING]
>
> **Preview**
>
>
> This feature is
>
> subject to the "Pre-GA Offerings Terms" in the General Service Terms section of the
> [Service Specific
> Terms](https://docs.cloud.google.com/terms/service-terms#1).
>
> Pre-GA features are available "as is" and might have limited support.
>
> For more information, see the
> [launch stage descriptions](https://cloud.google.com/products/#product-launch-stages).

Provisioned Throughput supports
[implicit caching](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-overview.md.txt#implicit-caching)
and [explicit caching](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-overview.md.txt#explicit-caching).
For information about supported models and limits, see [Context
caching](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-overview).

Implicit caching is enabled in all Google Cloud projects, by default. Implicit
caching reduces cost and latency during cache hits. Cached
tokens are charged at a [discount](https://cloud.google.com/gemini-enterprise-agent-platform/generative-ai/pricing)
relative to standard input tokens when a cache hit occurs.
Explicit caching offers more control and ensures a discount when explicit caches
are referenced. To learn how to create a context cache for explicit caching,
see [Create a context cache](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-create.md.txt).
To view model-specific discounts, see
[Context caching overview](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/context-cache/context-cache-overview.md.txt).
For Provisioned Throughput, the discount is applied through a
reduced burndown rate.

For example, Gemini 2.5 Pro has the following burndown rates for input
text tokens and cached tokens:

- 1 input text token = 1 token

- 1 input cached text token = 0.1 tokens

Sending 1,000 input tokens to this model results in a burndown of your
Provisioned Throughput by 1,000 input tokens per second. However,
if you send 1,000 cached tokens to Gemini 2.5 Pro, this results in a
burndown of your Provisioned Throughput by 100 tokens per second.

Note that this can lead to higher throughput for similar queries where the tokens
aren't cached and the cache discount isn't applied.

To view the burndown rates for models supported in Provisioned Throughput,
see [Supported models and burndown rates](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models.md.txt).

## Understand the burndown for Gemini Live API

**Request access:** The model `gemini-live-2.5-flash` is in private GA. For information about access to this release, see the [access request page](https://docs.google.com/forms/d/e/1FAIpQLScxBeD4UJ8GbUfX4SXjj5a1XJ1K7Urwvb0iSGdGccNcFRBrpQ/viewform).

Provisioned Throughput supports Gemini 3.8 Live and
Gemini 2.5 Flash with Gemini Live API. To understand how to calculate
the burndown when using Gemini Live API, see
[Provisioned Throughput for Gemini Live API](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/live-api.md.txt)
(and [Estimating GSU requirements](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/live-api.md.txt#estimating-gsu-requirements)
for Gemini 3.8 Live).

## Example of estimating your Provisioned Throughput needs

To estimate your Provisioned Throughput needs, use the
[estimation tool in the Google Cloud console](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput.md.txt#estimate-provisioned-throughput).
The following example illustrates the process of estimating the amount of
Provisioned Throughput for your model. The region isn't considered
in the estimation calculations.

This table provides the burndown rates for Gemini 3.8 Flash that you
can use to follow the example.

| **Model** | **Throughput per GSU** | **Units** | **Minimum GSU purchase increment** | **Burndown rates** |
|---|---|---|---|---|
| Gemini 3.8 Flash | 675 | Tokens | 1 | 1 input text token = 1 token 1 input image token = 1 token 1 input video token = 1 token 1 input audio token = 1 token 1 input text cache read token = 0.1 tokens 1 input image cache read token = 0.1 tokens 1 input video cache read token = 0.1 tokens 1 input audio cache read token = 0.1 tokens 1 output text response token = 5 tokens |

1. **Gather your requirements.** In this example, your requirement is to verify
   that you can support 10 queries per second (QPS) of a query with an
   input of 1,000 text tokens and 500 audio tokens, to receive an
   output of 300 text response tokens using `gemini-3.8-flash`.

   To calculate your throughput for your specific use case, use the [burndown
   rates](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/supported-models#google-models)
   for your selected model.
2. **Calculate your throughput.** Multiply your inputs by the burndown rates to
   arrive at total input tokens:

   $\\text{1,000} \\times (\\text{1 token per input text token}) + \\text{500}
   \\times (\\text{1 token per input audio token}) = \\text{1,500}$ burndown
   adjusted input tokens per query.

   Multiply your outputs by the burndown rates to arrive at total output
   tokens:

   $\\text{300} \\times (\\text{5 tokens per output text response token}) =
   \\text{1,500}$ burndown adjusted output tokens per query.

   Add your totals together:

   $\\text{1,500 burndown adjusted input tokens} + \\text{1,500 burndown adjusted
   output tokens} = \\text{3,000}$ total tokens per query.

   Multiply the total number of tokens by the QPS to arrive at total throughput
   per second:

   $\\text{3,000 total tokens per query} \\times \\text{10 QPS} = \\text{30,000}$
   total tokens per second.
3. **Calculate your needed GSUs.** The GSUs are the total tokens per second
   divided by the per-second throughput per GSU from the burndown table:

   $\\text{30,000 total tokens per second} \\div \\text{675 per-second throughput
   per GSU} = \\text{44.44}$ GSUs.

   The minimum GSU purchase increment for `gemini-3.8-flash` is **1** , so
   you'll need **45 GSUs** to assure your workload.

## What's next

- [Purchase Provisioned Throughput](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/provisioned-throughput/purchase-provisioned-throughput.md.txt).