Flexible Savings Plans (FSPs) are a type of spend-based committed use discount
(CUD) that help you save money on your usage on Google Cloud. With an FSP,
you commit to a minimum spend amount on a collection of eligible products
(including first-party, open-source, and participating third-party models) for a
monthly entitlement window, for the duration of the subscription term (1 year or
3 years). In exchange for your commitment, you receive discounts on eligible
usage.

FSPs introduce two new features to the Spend-Based CUDs:

- Subscriptions as part of the [eligible SKUs](https://docs.cloud.google.com/docs/cuds-flexible-savings-plans.md.txt#find_skus).
- Monthly [entitlement windows](https://docs.cloud.google.com/docs/cuds-flexible-savings-plans.md.txt#entitlement-windows)

Unlike standard spend-based [committed use discounts (CUDs)](https://docs.cloud.google.com/docs/cuds.md.txt) that
use an hourly entitlement window, FSPs use a **monthly** entitlement window.
This makes them ideal for workloads with spiky or unpredictable usage patterns.

## Find eligible SKUs

Your Gemini Enterprise FSP applies to eligible Google Cloud generative
AI models and product offerings, including first-party models (such as
Gemini), open-source Model-as-a-Service offerings, and
participating third-party (3P) foundation models available through
Vertex AI and Cloud Marketplace. To see the complete list of
SKUs in scope, see the [Gemini Enterprise Flexible Savings Plan
Eligible
SKUs](https://cloud.google.com/skus/sku-groups/gemini-enterprise-flexible-savings-plan-eligible-skus)
page.

## How Flexible Savings Plans work

When you purchase a Flexible Savings Plan, you commit to a specific monthly
spend amount for a 1-year or 3-year term. The system monitors your usage and
applies the [discounted applicable consumption model
rates](https://docs.cloud.google.com/docs/cuds-multiprice).

You won't be billed for your FSP if you used your monthly commitment amount on
eligible products. Your FSP commitment draws down as you use eligible products.

> [!IMPORTANT]
> **Important:** You **cannot cancel** the commitments you have purchased. If you accidentally purchased a commitment or made a mistake configuring your commitment, contact **Cloud Billing support** for help.

### Scope and pricing

The following rules apply to FSP scope and pricing:

- **Subscription term limit**: Subscriptions with a term greater than 1 month can't be discounted by an FSP. For example, if you have a 1-year subscription to a product, the FSP discount doesn't apply to it.
- **Pay the better rate** : For all eligible products and subscriptions, you always pay the lower (better) rate. If you have a negotiated price for any product within this commitment, the lowest price will apply. An FSP never increases the price of any product beyond its standard on-demand rate. For details, see [Pay the better rate](https://docs.cloud.google.com/docs/cuds-flexible-savings-plans.md.txt#pay-the-better-rate).
- **Discount-ineligible products** : Some products don't get FSP discounts but still draw down from the FSP. For those prices, see the [pricing table report](https://docs.cloud.google.com/billing/docs/how-to/pricing-table.md.txt).

#### Gemini Enterprise pricing

A 1-year Flexible Savings Plan provides a 10% discount and a 3-year Flexible
Savings Plan provides a 20% discount for all SKUs in the [Gemini Enterprise Flexible Savings Plan Eligible
SKUs](https://cloud.google.com/skus/sku-groups/gemini-enterprise-flexible-savings-plan-eligible-skus)
page, other than the following SKUs which don't get any incremental discounts
from the FSP, but still draw down the FSP commitment:

> [!NOTE]
> **Note:** Third-party products don't get any discounts from the FSP. Go to the [Pricing page](https://console.cloud.google.com/billing/pricing) in the Google Cloud console to view your final pricing.

| Product | Description | SKU ID |
|---|---|---|
| Provisioned Throughput | Vertex AI: Provisioned Throughput 3 month | `13D9-6212-6E17` |
| Provisioned Throughput | Vertex AI: Provisioned Throughput 1 Year | `9192-74A6-DE45` |
| Provisioned Throughput | Agent Platform Models: Provisioned Throughput 3 month Non Global | `F878-80F5-63BF` |
| Provisioned Throughput | Agent Platform Models: Provisioned Throughput 1 Year NON GLOBAL | `2106-EA10-F9CC` |
| Provisioned Throughput | Vertex AI Open Source Model as a Service: Provisioned Throughput 3 months | `1E0D-E522-0E12` |
| Provisioned Throughput | Vertex AI Open Source Model as a Service: Provisioned Throughput 1 year | `6009-4CB8-85D6` |
| Full Fine-Tuning | 1P foundational models full fine tuning service 1 year commit duration | `88CB-4A51-012D` |
| Full Fine-Tuning | 1P foundational models full fine tuning service 3 year commit duration | `0280-9537-E40B` |
| Full Fine-Tuning | Inferencing Full Fine Tuned 1P models 1 year commitment | `74E1-4528-1F94` |
| Full Fine-Tuning | Inferencing Full Fine Tuned 1P models 3 year commitment | `26B4-6E81-10C0` |
| Speech-to-Text | Speech-to-Text On-Device Language Subscription | `9886-4062-830D` |
| Speech-to-Text | Speech-to-Text On-Device Serving Library Subscription | `BEF8-431D-9BF4` |
| Text-to-Speech | Text-to-Speech On-Device Serving Library Subscription | `3D6B-450F-BA51` |
| Text-to-Speech | Text-to-Speech On-Device Voice Subscription | `80AF-473F-9A8D` |
| Gemini Enterprise App | Gemini Business: Subscription - one month term | `D03F-9D37-4753` |
| Gemini Enterprise App | Gemini Business: Subscription - one year term | `FE0D-41B2-9575` |
| Gemini Enterprise App | Gemini Enterprise Business Edition: Subscription Prepay - one month term | `FA2D-55B2-023F` |
| Gemini Enterprise App | Gemini Enterprise Business Edition: Subscription Prepay - one year term | `FB43-B23B-BD15` |
| Gemini Enterprise App | Gemini Enterprise Standard: Subscription - one month term | `0532-C2F0-1DF0` |
| Gemini Enterprise App | Gemini Enterprise Standard: Subscription - one year term | `B3D3-B9E1-610F` |
| Gemini Enterprise App | Gemini Enterprise Plus: Subscription - one month term | `4EDF-A125-F89E` |
| Gemini Enterprise App | Gemini Enterprise Plus: Subscription - one year term | `5657-9AD2-26F6` |
| Gemini Enterprise App | Gemini Enterprise Frontline: Subscription - one month term | `C3B7-10E2-45EE` |
| Gemini Enterprise App | Gemini Enterprise Frontline: Subscription - one year term | `070B-47E2-47B3` |
| Gemini Enterprise App | Gemini Enterprise Standard Emerging: Subscription - one month term | `54F6-590C-84B5` |
| Gemini Enterprise App | Gemini Enterprise Standard Emerging: Subscription - one year term | `C853-D67D-D911` |
| Gemini Enterprise App | Gemini Notebook Enterprise: Subscription - one month term | `BF82-0F2F-7FA3` |
| Gemini Enterprise App | Gemini Notebook Enterprise: Subscription - one year term | `8597-66F8-5FF1` |

#### Third-party models and Private Offers

Gemini Enterprise Flexible Savings Plans support participating
third-party (3P) foundation models offered through Vertex AI and
Cloud Marketplace.

##### Interactions with Private Offers

The following rules govern how FSPs and Private Offers interact:

- **Benefit application hierarchy** :
  1. **Private Offer Discounts**: If you purchased an FSP-eligible 3P product through a Private Offer that gives you discounts on the partners list price, the discounted prices and Private Offer terms will apply. Additionally, your discounted spend will draw down your monthly FSP commitment.
  2. **Private Offer CUD credits**: If you purchased an FSP-eligible 3P product through a Private Offer that gives you credits, the purchase of or installments for the Private Offer will draw down your monthly FSP commitment. Note that your usage of your purchased credits won't additionally draw down your monthly FSP commitment.

##### Enterprise Agreement commitment drawdown

Spend on third-party models under an FSP draws down against your organization's
enterprise commitment, subject to the standard Cloud Marketplace
drawdown rules and contract caps defined in your enterprise agreement.

### Entitlement windows

The entitlement window is the recurring period over which your usage is measured
against your commitment amount. If you don't use your entire commitment amount
within the entitlement window, you will billed for the remaining amount.
Underutilized FSP commitments don't carry forward.

FSPs support a **monthly** entitlement window. This means you have the calendar
month to consume your committed spend. However, the billing system evaluates
your usage and applies the FSP discounts on an hourly basis.

For example, if you have a $100 monthly commitment, the system checks your usage
every hour. If you have eligible usage in hour 1, the discount is applied and
counts toward your $100 commitment. This continues throughout the month until
you reach your $100 limit or the month ends. If you exceed $100 of usage before
the end of the month, subsequent usage is charged at the default consumption
model rate.

### Overutilized commitments (exceeding your commitment)

If your eligible usage in a given month exceeds the monthly amount you committed
to, the FSP covers usage up to your limit, and the remainder is billed at either
the standard on-demand rate or your negotiated price for any product within this
commitment.

- **Mechanism**: The system applies the FSP discounted rate to the usage equal to your monthly commitment amount.
- **Cost** : Any usage *beyond* that limit in the month is charged at either the standard **on-demand price** (the "Default" consumption model) or your negotiated price for any product within the commitment. You don't receive any FSP discount on the overage.

### Underutilized commitments (not meeting your commitment)

If your eligible usage in a given month is less than your committed amount, you
are still responsible for the full cost of the commitment.

- **No rollover** : Unused commitment does *NOT* carry over to the next month.
- **Cost**: You pay for the full monthly commitment amount you purchased. You are charged a commitment fee for the unused portion of the commitment at the end of the month.

### How discounts are applied

Within an FSP's scope, discounts are applied in the following order:

1. **Usage time**: Costs in an earlier usage hour are applied first.
2. **Discount level**: Usage with the steepest discount percentage relative to the on-demand list price within a given usage hour is applied next. This greedy selection ensures that your commitment dollars deliver the highest savings rate first.
3. **Proportional application**: If usage across multiple resources shares the same discount percentage within an hour, the commitment is applied proportionally to all costs within that discount level and usage hour.

If you have a subscription that is covered by an FSP, the subscription benefits
apply to your usage first. Your subscription cost counts toward meeting your FSP
spend commitment, but the FSP doesn't apply an additional discount to the
subscription price.

#### Pay the better rate

The lower of the default consumption model price, 3P Private Offers, and the FSP
consumption model price is applied.

### Start times and proration

After you purchase an FSP, the commitment activates based on the time of
purchase:

- **Purchased at :00 to :49 (first 50 minutes of the hour)**: Starts at the beginning of the next hour.
- **Purchased at :50 to :59 (last 10 minutes of the hour)**: Skips an hour and starts at the beginning of the hour after next.

All activation and entitlement window boundaries operate in
**Pacific Time (PT)** (`America/Los_Angeles`).

#### Monthly proration for mid-month purchases

When you purchase an FSP mid-month with immediate activation, your initial
entitlement period is prorated for the remainder of that calendar month:

1. **Prorated initial month**: You receive a prorated spend entitlement for the period from your activation timestamp through the end of the current calendar month (11:59:59 PM PT on the last day of the month). Your commitment fee and underutilization threshold for that partial month are scaled proportionally to the fraction of elapsed hours in the month.
2. **Subsequent calendar months**: Starting on the first day of the next calendar month at 12:00 AM PT, each entitlement period aligns with a full calendar month for the remainder of your 1-year or 3-year term
3. **Prorated last month**: You receive a prorated spend entitlement for the period from the start of the current calendar month to the date your term ends. Your commitment fee and underutilization threshold for that partial month are scaled proportionally to the fraction of elapsed hours in the month.

For example, suppose you purchase a 1-year FSP with a $3,000 monthly commitment
that activates at 12:00 AM PT on April 16 (where April has 30 days total,
leaving 15 days remaining). Your commitment for April is prorated to $1,500
($3,000 × 15/30).

### Cancellations, migrations, and corrections

After purchase, you **cannot cancel** or modify an FSP. You must pay the
agreed-upon monthly amount for the duration of the commitment. If you
accidentally purchased a commitment or made a mistake configuring your
commitment, contact [Cloud Billing support](https://docs.cloud.google.com/billing/docs/support) for
help.

> [!NOTE]
> **Note:** Any corrections, modifications, or cancellations to your Flexible Savings Plan processed by Cloud Billing support only take effect at month boundaries.

### Attribution of discounts and fees

FSPs use proportional attribution, which is the same attribution model used for
other spend-based committed use discounts (CUDs). Costs are attributed every
hour. Fees are applied to the projects in your Cloud Billing account in
proportion to the amount of total eligible usage consumed by each project. For
details, see [Attribution of committed use discount fees and
credits](https://docs.cloud.google.com/docs/cuds-attribution).

## Purchase a Flexible Savings Plan

You can purchase FSPs directly [using the Google Cloud console
(self-service)](https://docs.cloud.google.com/docs/cuds-multiprice#steps_to_purchase).

To purchase a Flexible Savings Plan, you need the offer ID associated with the
commitment term. To find offer IDs, see [Identify the Offer you want to
purchase](https://docs.cloud.google.com/marketplace/docs/commitment-api-purchasing).

When configuring your FSP, you specify:

- The **subscription term** (1 year or 3 years).
- The **commitment amount** per monthly entitlement window.
- A **name** for your FSP. This name must be unique across all your FSPs and committed use discounts (CUDs).

## View your Flexible Savings Plans

You can view your active and historical Flexible Savings Plans on the
[**Committed use discounts (CUDs)** page](https://console.cloud.google.com/billing/reports/commitments)
in the Google Cloud console.

The page provides two levels of detail (**Aggregated view** and **Instance
view** ) to help you manage your commitments. For more information about these
views, see [View your commitments](https://docs.cloud.google.com/billing/docs/how-to/cuds-list-overview.md.txt).

From the **Instance view**, you can see the breakdown of entitlement windows.

## Monitor and analyze your Flexible Savings Plans

You can monitor and manage your FSPs using standard Cloud Billing
FinOps tools:

- **Cost reports**: View your FSP commitment fees and on-demand costs. You can isolate costs for fees and usage covered by your FSPs.
- **Cost table report**: Reconcile your invoice by viewing the detailed breakdown of your FSP charges.
- **Cloud Billing export to BigQuery** : [Set up Cloud Billing data export to
  BigQuery](https://docs.cloud.google.com/billing/docs/how-to/export-data-bigquery-setup) to programmatically track your active FSPs, historical performance, and utilization metrics. You can use the [sample queries](https://docs.cloud.google.com/docs/cuds-flexible-savings-plans.md.txt#sample-queries) to analyze this data.

## Sample queries

Use the following sample queries to analyze your Flexible Savings Plans (FSPs)
by using your Cloud Billing export to BigQuery data.

### Calculate commitment savings

Use the following queries to calculate your commitment savings in dollars and
as a percentage for open and closed entitlement windows.

#### Open window

This query calculates gross savings and the gross savings rate for the
entitlement window that is open: on-demand cost minus the FSP-discounted cost
of the usage that the FSP covered. Underutilization fees aren't included
because they're charged only after the window closes. To filter for specific
consumption models, replace `'%'` in `LIKE ANY ('%')` with the target
consumption model IDs.

```
WITH
  fsp_windows AS (
    SELECT DISTINCT
      IF(
        STARTS_WITH(subscription.instance_id, 'subscriptions/'),
        subscription.instance_id,
        CONCAT('subscriptions/', subscription.instance_id)) AS fsp_instance_id,
      subscription.display_name AS fsp_display_name,
      subscription.entitlement_period,
      consumption_model.id AS consumption_model_id,
      consumption_model.description AS consumption_model_description,
      entitlement_window.start_time AS window_start_time,
      entitlement_window.end_time AS window_end_time
    FROM `PROJECT.DATASET.cud_subscriptions_export`,
      UNNEST(entitlement_schedule) AS entitlement_window
    WHERE
      DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME)) FROM `PROJECT.DATASET.cud_subscriptions_export`)
      AND consumption_model.id LIKE ANY ('%')
      AND CURRENT_TIMESTAMP() >= entitlement_window.start_time
      AND CURRENT_TIMESTAMP() < entitlement_window.end_time
  ),
  fsp_cost AS (
    SELECT
      COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) AS fsp_instance_id,
      IFNULL(consumption_model.applied_subscription_instance_id, '') = '' AS is_fee_line,
      usage_start_time,
      currency,
      SUM(CAST(cost AS NUMERIC)) AS cost,
      SUM(CAST(cost_at_effective_price_default AS NUMERIC)) AS cost_at_on_demand_price
    FROM `PROJECT.DATASET.gcp_billing_export_resource_v1_BILLING_ACCOUNT_ID`
    WHERE
      DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE('America/Los_Angeles'), INTERVAL 93 DAY)
      AND COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) IN (SELECT fsp_instance_id FROM fsp_windows)
    GROUP BY ALL
  ),
  fsp_window_totals AS (
    SELECT
      fsp_windows.fsp_instance_id,
      fsp_windows.fsp_display_name,
      fsp_windows.entitlement_period,
      fsp_windows.consumption_model_id,
      fsp_windows.consumption_model_description,
      fsp_windows.window_start_time,
      fsp_windows.window_end_time,
      fsp_cost.currency,
      IFNULL(SUM(IF(fsp_cost.is_fee_line, 0, fsp_cost.cost_at_on_demand_price)), 0)
        AS on_demand_cost,
      IFNULL(SUM(IF(fsp_cost.is_fee_line, 0, fsp_cost.cost)), 0) AS usage_cost
    FROM fsp_windows
    LEFT JOIN fsp_cost
      ON
        fsp_cost.fsp_instance_id = fsp_windows.fsp_instance_id
        AND fsp_cost.usage_start_time >= fsp_windows.window_start_time
        AND fsp_cost.usage_start_time < fsp_windows.window_end_time
    GROUP BY ALL
  )
SELECT
  fsp_instance_id,
  fsp_display_name,
  entitlement_period,
  consumption_model_id,
  consumption_model_description,
  DATETIME(window_start_time, 'America/Los_Angeles') AS window_start_time_pt,
  DATETIME(window_end_time, 'America/Los_Angeles') AS window_end_time_pt,
  currency,
  on_demand_cost - usage_cost AS gross_commitment_savings,
  SAFE_DIVIDE(on_demand_cost - usage_cost, on_demand_cost) AS gross_commitment_savings_rate
FROM fsp_window_totals
ORDER BY fsp_instance_id, window_start_time_pt
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.
- `BILLING_ACCOUNT_ID`: Your Cloud Billing Account API ID, with hyphens replaced by underscores.

#### Closed window

This query calculates net commitment savings and the net savings rate for
completed entitlement windows after underutilization fees
(`FEE_UTILIZATION_OFFSET`) are applied. In `fsp_windows` and `fsp_cost`, update
`DATE '2026-07-01'` and `DATE '2026-10-01'` to match your target window range,
and replace `'%'` in `LIKE ANY ('%')` if you want to filter for specific
consumption models.

```
WITH
  fsp_windows AS (
    SELECT DISTINCT
      IF(
        STARTS_WITH(subscription.instance_id, 'subscriptions/'),
        subscription.instance_id,
        CONCAT('subscriptions/', subscription.instance_id)) AS fsp_instance_id,
      subscription.display_name AS fsp_display_name,
      subscription.entitlement_period,
      consumption_model.id AS consumption_model_id,
      consumption_model.description AS consumption_model_description,
      entitlement_window.start_time AS window_start_time,
      entitlement_window.end_time AS window_end_time
    FROM `PROJECT.DATASET.cud_subscriptions_export`,
      UNNEST(entitlement_schedule) AS entitlement_window
    WHERE
      DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME)) FROM `PROJECT.DATASET.cud_subscriptions_export`)
      AND consumption_model.id LIKE ANY ('%')
      AND entitlement_window.start_time >= TIMESTAMP(DATE '2026-07-01', 'America/Los_Angeles')
      AND entitlement_window.end_time <= TIMESTAMP(DATE '2026-10-01', 'America/Los_Angeles')
      AND entitlement_window.end_time <= CURRENT_TIMESTAMP()
  ),
  fsp_cost AS (
    SELECT
      COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) AS fsp_instance_id,
      IFNULL(consumption_model.applied_subscription_instance_id, '') = '' AS is_fee_line,
      usage_start_time,
      currency,
      SUM(CAST(cost AS NUMERIC)) AS cost,
      SUM(CAST(cost_at_effective_price_default AS NUMERIC)) AS cost_at_on_demand_price,
      SUM(
        (
          SELECT SUM(CAST(credit.amount AS NUMERIC))
          FROM UNNEST(credits) AS credit
          WHERE credit.type = 'FEE_UTILIZATION_OFFSET'
        )) AS fee_utilization_offset
    FROM `PROJECT.DATASET.gcp_billing_export_resource_v1_BILLING_ACCOUNT_ID`
    WHERE
      DATE(_PARTITIONTIME) >= DATE '2026-07-01'
      AND COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) IN (SELECT fsp_instance_id FROM fsp_windows)
    GROUP BY ALL
  ),
  fsp_window_totals AS (
    SELECT
      fsp_windows.fsp_instance_id,
      fsp_windows.fsp_display_name,
      fsp_windows.entitlement_period,
      fsp_windows.consumption_model_id,
      fsp_windows.consumption_model_description,
      fsp_windows.window_start_time,
      fsp_windows.window_end_time,
      fsp_cost.currency,
      IFNULL(SUM(IF(fsp_cost.is_fee_line, 0, fsp_cost.cost_at_on_demand_price)), 0)
        AS on_demand_cost,
      IFNULL(SUM(fsp_cost.cost + IFNULL(fsp_cost.fee_utilization_offset, 0)), 0) AS actual_fsp_cost
    FROM fsp_windows
    LEFT JOIN fsp_cost
      ON
        fsp_cost.fsp_instance_id = fsp_windows.fsp_instance_id
        AND fsp_cost.usage_start_time >= fsp_windows.window_start_time
        AND fsp_cost.usage_start_time < fsp_windows.window_end_time
    GROUP BY ALL
  )
SELECT
  fsp_instance_id,
  fsp_display_name,
  entitlement_period,
  consumption_model_id,
  consumption_model_description,
  DATETIME(window_start_time, 'America/Los_Angeles') AS window_start_time_pt,
  DATETIME(window_end_time, 'America/Los_Angeles') AS window_end_time_pt,
  currency,
  on_demand_cost - actual_fsp_cost AS net_commitment_savings,
  SAFE_DIVIDE(on_demand_cost - actual_fsp_cost, on_demand_cost) AS net_commitment_savings_rate
FROM fsp_window_totals
ORDER BY fsp_instance_id, window_start_time_pt
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.
- `BILLING_ACCOUNT_ID`: Your Cloud Billing Account API ID, with hyphens replaced by underscores.

### Calculate commitment utilization

Commitment utilization measures how effectively you use your commitments.

#### Open window

This query calculates in-progress utilization for the entitlement window that
is open: FSP-covered usage so far at FSP prices in USD divided by the
commitment amount for the window. To filter for specific consumption models,
replace `'%'` in `LIKE ANY ('%')` with the target consumption model IDs.

```
WITH
  fsp_windows AS (
    SELECT DISTINCT
      IF(
        STARTS_WITH(subscription.instance_id, 'subscriptions/'),
        subscription.instance_id,
        CONCAT('subscriptions/', subscription.instance_id)) AS fsp_instance_id,
      subscription.display_name AS fsp_display_name,
      subscription.entitlement_period,
      consumption_model.id AS consumption_model_id,
      consumption_model.description AS consumption_model_description,
      entitlement_window.start_time AS window_start_time,
      entitlement_window.end_time AS window_end_time,
      entitlement_window.commitment_amount AS window_commitment_amount_usd
    FROM `PROJECT.DATASET.cud_subscriptions_export`,
      UNNEST(entitlement_schedule) AS entitlement_window
    WHERE
      DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME)) FROM `PROJECT.DATASET.cud_subscriptions_export`)
      AND consumption_model.id LIKE ANY ('%')
      AND CURRENT_TIMESTAMP() >= entitlement_window.start_time
      AND CURRENT_TIMESTAMP() < entitlement_window.end_time
  ),
  fsp_cost AS (
    SELECT
      COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) AS fsp_instance_id,
      IFNULL(consumption_model.applied_subscription_instance_id, '') = '' AS is_fee_line,
      usage_start_time,
      SUM(CAST(cost AS NUMERIC) / CAST(currency_conversion_rate AS NUMERIC)) AS cost_usd
    FROM `PROJECT.DATASET.gcp_billing_export_resource_v1_BILLING_ACCOUNT_ID`
    WHERE
      DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE('America/Los_Angeles'), INTERVAL 93 DAY)
      AND COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) IN (SELECT fsp_instance_id FROM fsp_windows)
    GROUP BY ALL
  ),
  fsp_window_totals AS (
    SELECT
      fsp_windows.fsp_instance_id,
      fsp_windows.fsp_display_name,
      fsp_windows.entitlement_period,
      fsp_windows.consumption_model_id,
      fsp_windows.consumption_model_description,
      fsp_windows.window_start_time,
      fsp_windows.window_end_time,
      fsp_windows.window_commitment_amount_usd,
      IFNULL(SUM(IF(fsp_cost.is_fee_line, 0, fsp_cost.cost_usd)), 0) AS usage_cost_usd
    FROM fsp_windows
    LEFT JOIN fsp_cost
      ON
        fsp_cost.fsp_instance_id = fsp_windows.fsp_instance_id
        AND fsp_cost.usage_start_time >= fsp_windows.window_start_time
        AND fsp_cost.usage_start_time < fsp_windows.window_end_time
    GROUP BY ALL
  )
SELECT
  fsp_instance_id,
  fsp_display_name,
  entitlement_period,
  consumption_model_id,
  consumption_model_description,
  DATETIME(window_start_time, 'America/Los_Angeles') AS window_start_time_pt,
  DATETIME(window_end_time, 'America/Los_Angeles') AS window_end_time_pt,
  window_commitment_amount_usd,
  usage_cost_usd AS covered_usage_cost_usd,
  SAFE_DIVIDE(usage_cost_usd, window_commitment_amount_usd) AS commitment_utilization_rate
FROM fsp_window_totals
ORDER BY fsp_instance_id, window_start_time_pt
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.
- `BILLING_ACCOUNT_ID`: Your Cloud Billing Account API ID, with hyphens replaced by underscores.

#### Closed window

This query calculates final utilization for entitlement windows that have ended
from the FSP fee line item: the `FEE_UTILIZATION_OFFSET` credit, which
represents the utilized part of the commitment, divided by the commitment
amount for the window. In `fsp_windows` and `fsp_cost`, update
`DATE '2026-07-01'` and `DATE '2026-10-01'` to match your target window range,
and replace `'%'` in `LIKE ANY ('%')` if you want to filter for specific
consumption models.

```
WITH
  fsp_windows AS (
    SELECT DISTINCT
      IF(
        STARTS_WITH(subscription.instance_id, 'subscriptions/'),
        subscription.instance_id,
        CONCAT('subscriptions/', subscription.instance_id)) AS fsp_instance_id,
      subscription.display_name AS fsp_display_name,
      subscription.entitlement_period,
      consumption_model.id AS consumption_model_id,
      consumption_model.description AS consumption_model_description,
      entitlement_window.start_time AS window_start_time,
      entitlement_window.end_time AS window_end_time,
      entitlement_window.commitment_amount AS window_commitment_amount_usd
    FROM `PROJECT.DATASET.cud_subscriptions_export`,
      UNNEST(entitlement_schedule) AS entitlement_window
    WHERE
      DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME)) FROM `PROJECT.DATASET.cud_subscriptions_export`)
      AND consumption_model.id LIKE ANY ('%')
      AND entitlement_window.start_time >= TIMESTAMP(DATE '2026-07-01', 'America/Los_Angeles')
      AND entitlement_window.end_time <= TIMESTAMP(DATE '2026-10-01', 'America/Los_Angeles')
      AND entitlement_window.end_time <= CURRENT_TIMESTAMP()
  ),
  fsp_cost AS (
    SELECT
      COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) AS fsp_instance_id,
      usage_start_time,
      SUM(
        (
          SELECT SUM(CAST(credit.amount AS NUMERIC))
          FROM UNNEST(credits) AS credit
          WHERE credit.type = 'FEE_UTILIZATION_OFFSET'
        )
        / CAST(currency_conversion_rate AS NUMERIC)) AS fee_utilization_offset_usd
    FROM `PROJECT.DATASET.gcp_billing_export_resource_v1_BILLING_ACCOUNT_ID`
    WHERE
      DATE(_PARTITIONTIME) >= DATE '2026-07-01'
      AND COALESCE(
        NULLIF(consumption_model.applied_subscription_instance_id, ''),
        subscription.instance_id) IN (SELECT fsp_instance_id FROM fsp_windows)
    GROUP BY ALL
  ),
  fsp_window_totals AS (
    SELECT
      fsp_windows.fsp_instance_id,
      fsp_windows.fsp_display_name,
      fsp_windows.entitlement_period,
      fsp_windows.consumption_model_id,
      fsp_windows.consumption_model_description,
      fsp_windows.window_start_time,
      fsp_windows.window_end_time,
      fsp_windows.window_commitment_amount_usd,
      IFNULL(SUM(fsp_cost.fee_utilization_offset_usd), 0) AS fee_utilization_offset_usd
    FROM fsp_windows
    LEFT JOIN fsp_cost
      ON
        fsp_cost.fsp_instance_id = fsp_windows.fsp_instance_id
        AND fsp_cost.usage_start_time >= fsp_windows.window_start_time
        AND fsp_cost.usage_start_time < fsp_windows.window_end_time
    GROUP BY ALL
  )
SELECT
  fsp_instance_id,
  fsp_display_name,
  entitlement_period,
  consumption_model_id,
  consumption_model_description,
  DATETIME(window_start_time, 'America/Los_Angeles') AS window_start_time_pt,
  DATETIME(window_end_time, 'America/Los_Angeles') AS window_end_time_pt,
  window_commitment_amount_usd,
  ABS(fee_utilization_offset_usd) AS utilized_commitment_usd,
  SAFE_DIVIDE(ABS(fee_utilization_offset_usd), window_commitment_amount_usd)
    AS commitment_utilization_rate
FROM fsp_window_totals
ORDER BY fsp_instance_id, window_start_time_pt
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.
- `BILLING_ACCOUNT_ID`: Your Cloud Billing Account API ID, with hyphens replaced by underscores.

### Size your purchase amount

Use closed, completed periods only to size your purchase amount. Open periods
usually contain little uncovered cost and can skew the result. The sizing query
evaluates historical on-demand usage that no active commitment covered during
those periods. Because historical usage doesn't reflect scheduled commitment
ramps, future activations, or expiring commitments, factor in your upcoming
entitlement schedules when sizing a purchase.

#### Find eligible SKUs and discount rates

This query returns FSP-eligible SKUs, their on-demand and FSP prices per
pricing tier, and the FSP discount rate so that you can validate the prices
that the next query applies. SKUs with a 0% discount, such as
Provisioned Throughput, are kept because they still draw down the
commitment. Every consumption model whose name contains "Flexible Savings Plan"
is included. To include only specific consumption models, replace `'%'` in
`LIKE ANY ('%')` with their IDs, such as only the quarterly 1-year model.

```
WITH
  fsp_tier_prices AS (
    SELECT
      pricing.sku.id AS sku_id,
      pricing.sku.description AS sku_description,
      consumption_model_price.consumption_model_id,
      consumption_model_price.consumption_model_display_name,
      on_demand_tier.start_usage_amount AS tier_start_usage_amount,
      CAST(
        COALESCE(
          (
            SELECT tier.usd_amount
            FROM UNNEST(pricing.billing_account_price.tiered_rates) AS tier
            WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
          ),
          on_demand_tier.usd_amount) AS NUMERIC) AS on_demand_price_usd,
      CAST(
        COALESCE(
          (
            SELECT tier.usd_amount
            FROM UNNEST(consumption_model_price.billing_account_price.tiered_rates) AS tier
            WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
          ),
          (
            SELECT tier.usd_amount
            FROM UNNEST(consumption_model_price.list_price.tiered_rates) AS tier
            WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
          )) AS NUMERIC) AS fsp_price_usd
    FROM `PROJECT.DATASET.cloud_pricing_export` AS pricing,
      UNNEST(pricing.consumption_model_prices) AS consumption_model_price,
      UNNEST(pricing.list_price.tiered_rates) AS on_demand_tier
    WHERE
      DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY)
      AND DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME))
        FROM `PROJECT.DATASET.cloud_pricing_export`
        WHERE DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY)
      )
      AND consumption_model_price.consumption_model_display_name LIKE '%Flexible Savings Plan%'
      AND consumption_model_price.consumption_model_id LIKE ANY ('%')
  )
SELECT
  sku_id,
  sku_description,
  consumption_model_id,
  consumption_model_display_name,
  tier_start_usage_amount,
  on_demand_price_usd,
  fsp_price_usd,
  SAFE_DIVIDE(on_demand_price_usd - fsp_price_usd, on_demand_price_usd) AS fsp_discount_rate
FROM fsp_tier_prices
ORDER BY consumption_model_display_name, sku_id, tier_start_usage_amount
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.

#### Apply discount rates to uncovered usage

This query applies each SKU's FSP price ratio per pricing tier, using the same
pricing logic as the preceding query, to uncovered on-demand usage in each past
period. It returns the potential FSP cost per period and a suggested commitment
per window based on the lowest stable period, also called the low watermark.
Running the preceding query isn't a prerequisite because the pricing step is
built in.

Before running the query, edit the following inputs in place in `sizing_periods`
and `uncovered_cost`:

- **Consumption models** : In `fsp_tier_price_ratios`, choose consumption models in `LIKE ANY ('%')` that match the period length, such as the quarterly FSP consumption models with 3-month periods.
- **Period length** : In `sizing_periods`, use `1` for monthly FSPs and `3` for quarterly FSPs.
- **First period start** : Update `DATE '2026-07-01'`, which appears twice in `sizing_periods` and in the `_PARTITIONTIME` filter of `uncovered_cost`. Keep both dates the same.
- **Number of periods** : Update the `3` in `GENERATE_ARRAY(0, 3)`, such as `3` monthly periods or `2` to `4` quarterly periods. All periods must have ended.

```
WITH
  fsp_tier_price_ratios AS (
    SELECT
      pricing.sku.id AS sku_id,
      consumption_model_price.consumption_model_id,
      consumption_model_price.consumption_model_display_name,
      on_demand_tier.start_usage_amount AS tier_start_usage_amount,
      SAFE_DIVIDE(
        CAST(
          COALESCE(
            (
              SELECT tier.usd_amount
              FROM UNNEST(consumption_model_price.billing_account_price.tiered_rates) AS tier
              WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
            ),
            (
              SELECT tier.usd_amount
              FROM UNNEST(consumption_model_price.list_price.tiered_rates) AS tier
              WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
            )) AS NUMERIC),
        CAST(
          COALESCE(
            (
              SELECT tier.usd_amount
              FROM UNNEST(pricing.billing_account_price.tiered_rates) AS tier
              WHERE tier.start_usage_amount = on_demand_tier.start_usage_amount
            ),
            on_demand_tier.usd_amount) AS NUMERIC)) AS fsp_price_ratio
    FROM `PROJECT.DATASET.cloud_pricing_export` AS pricing,
      UNNEST(pricing.consumption_model_prices) AS consumption_model_price,
      UNNEST(pricing.list_price.tiered_rates) AS on_demand_tier
    WHERE
      DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY)
      AND DATE(_PARTITIONTIME) = (
        SELECT MAX(DATE(_PARTITIONTIME))
        FROM `PROJECT.DATASET.cloud_pricing_export`
        WHERE DATE(_PARTITIONTIME) >= DATE_SUB(CURRENT_DATE(), INTERVAL 7 DAY)
      )
      AND consumption_model_price.consumption_model_display_name LIKE '%Flexible Savings Plan%'
      AND consumption_model_price.consumption_model_id LIKE ANY ('%')
  ),
  sizing_periods AS (
    SELECT period_start_date, period_end_date
    FROM (
      SELECT
        period_start_date,
        LEAD(period_start_date) OVER (ORDER BY period_start_date) AS period_end_date
      FROM UNNEST(GENERATE_ARRAY(0, 3)) AS period_index,
        UNNEST([DATE_ADD(DATE '2026-07-01', INTERVAL period_index * 1 MONTH)]) AS period_start_date
    )
    WHERE period_end_date IS NOT NULL
  ),
  uncovered_cost AS (
    SELECT
      DATE(usage_start_time, 'America/Los_Angeles') AS usage_date,
      sku.id AS sku_id,
      IFNULL(price.tier_start_amount, 0) AS tier_start_usage_amount,
      SUM(CAST(cost AS NUMERIC) / CAST(currency_conversion_rate AS NUMERIC)) AS cost_usd
    FROM `PROJECT.DATASET.gcp_billing_export_resource_v1_BILLING_ACCOUNT_ID`
    WHERE
      DATE(_PARTITIONTIME) >= DATE '2026-07-01'
      AND IFNULL(consumption_model.applied_subscription_instance_id, '') = ''
      AND sku.id IN (SELECT sku_id FROM fsp_tier_price_ratios)
    GROUP BY ALL
  ),
  period_cost AS (
    SELECT
      fsp_tier_price_ratios.consumption_model_id,
      sizing_periods.period_start_date,
      SUM(uncovered_cost.cost_usd) AS uncovered_on_demand_cost_usd,
      SUM(uncovered_cost.cost_usd * fsp_tier_price_ratios.fsp_price_ratio) AS potential_fsp_cost_usd
    FROM uncovered_cost
    JOIN sizing_periods
      ON
        uncovered_cost.usage_date >= sizing_periods.period_start_date
        AND uncovered_cost.usage_date < sizing_periods.period_end_date
    JOIN fsp_tier_price_ratios
      ON
        fsp_tier_price_ratios.sku_id = uncovered_cost.sku_id
        AND fsp_tier_price_ratios.tier_start_usage_amount = uncovered_cost.tier_start_usage_amount
    GROUP BY ALL
  ),
  fsp_consumption_models AS (
    SELECT DISTINCT consumption_model_id, consumption_model_display_name
    FROM fsp_tier_price_ratios
  )
SELECT
  fsp_consumption_models.consumption_model_id,
  fsp_consumption_models.consumption_model_display_name,
  sizing_periods.period_start_date,
  sizing_periods.period_end_date,
  IFNULL(period_cost.uncovered_on_demand_cost_usd, 0) AS uncovered_on_demand_cost_usd,
  IFNULL(period_cost.potential_fsp_cost_usd, 0) AS potential_fsp_cost_usd,
  IFNULL(period_cost.uncovered_on_demand_cost_usd - period_cost.potential_fsp_cost_usd, 0)
    AS potential_savings_usd,
  MIN(IFNULL(period_cost.potential_fsp_cost_usd, 0))
    OVER (PARTITION BY fsp_consumption_models.consumption_model_id)
    AS suggested_commitment_per_window_usd
FROM fsp_consumption_models
CROSS JOIN sizing_periods
LEFT JOIN period_cost
  ON
    period_cost.consumption_model_id = fsp_consumption_models.consumption_model_id
    AND period_cost.period_start_date = sizing_periods.period_start_date
ORDER BY consumption_model_display_name, period_start_date
```

Replace the following:

- `PROJECT`: The ID of the Google Cloud project that contains your Cloud Billing export dataset.
- `DATASET`: The name of the BigQuery dataset containing your Cloud Billing export tables.
- `BILLING_ACCOUNT_ID`: Your Cloud Billing Account API ID, with hyphens replaced by underscores.