Using Agentforce and embedded AI features can impact credit consumption.
To use Flex Credits, Salesforce Foundations is required. You can get Salesforce Foundations in one of these ways.
- If you have access to Agentforce through an add-on license, activate Salesforce Foundations in your account. For step-by-step instructions, see Activate Foundations in your account. Requires Enterprise, Performance, Unlimited, or Developer Editions.
- If you access Agentforce through an Agentforce 1 Edition, Salesforce Foundations is included and active on your account.
To estimate the number of credits your org will use and calculate ROI based on that estimate, access the Agentforce ROI Estimator
To monitor your usage in these usage types, access your org’s Digital Wallet.
Flex Credits usage types as described in this article are based on standard measurement methods. However, some features can have specific limits or exceptions. Refer to your feature documentation for billing details unique to those features.
Actions Usage Types
Actions are the functions the AI agent uses to get information or perform tasks. There are two types of actions: standard and custom.
- Standard agent actions are actions that are available out of the box. See Standard Action Reference.
- Custom actions are actions that are created by you from scratch or by modifying a standard action. To learn more about what customer created actions are, see Create a Custom Action.
Agent actions are billed when they're run, whether they're run deterministically (for example, in reasoning instructions) or during agent reasoning with an LLM. Utilities (escalations to a representative, setting variables, or transitions to another topic) aren't billed as agent actions.
| Billing Category | Description |
|---|---|
| Standard Action |
Usage is determined by the number of standard actions called by the agent in a text conversation. Currently, every standard action called by the agent is metered as one standard action, irrespective of the number of tokens processed. Standard agent actions are actions that are available out-of-the-box. To check the list of Standard actions, see Standard Action Reference.  Note Use of some standard agent actions require that a subscription has been purchased for
each user that accesses these actions, such as a subscription to Agentforce for Sales
Add-on or Agentforce for Service Add-on. To determine which subscription is required for
such standard actions, see Standard Action Reference at Standard Action Reference. While this
requirement is not technically enforced yet, users who don’t have the required add-on
license will lose access to such actions for which they don’t have a license when the
requirement is enforced.
|
| Custom Action |
Usage is determined by the number of custom actions called by the agent in a text conversation. Currently, every custom action called by the agent is metered as one custom action, irrespective of the number of tokens processed. Async calls to Salesforce AI in custom action implementation may incur additional charges. |
| Standard Voice Action |
Usage is determined by the number of standard actions executed by the agent in a voice conversation. Note Use of some standard agent actions require that a subscription has been purchased for
each user that accesses these actions, such as a subscription to Agentforce for Sales
Add-on or Agentforce for Service Add-on. To determine which subscription is required for
such standard actions, see Standard Action Reference at Standard Action Reference. While this
requirement is not technically enforced yet, users who don’t have the required add-on
license will lose access to such actions for which they don’t have a license when the
requirement is enforced.
|
| Custom Voice Action |
Usage is determined by the number of custom actions called by the agent in a voice conversation. Custom actions are actions that are created by you or which result from your modification of a standard action. To learn more about what customer created actions are, see Create a Custom Agent Action. To learn more about how standard actions become custom actions, see Editing Standard Agent Action Reference Actions. Async calls to Salesforce AI in custom action implementation may incur additional charges. |
Help Agent Resolutions Usage Types
The pricing model for the Agentforce Help Agent is strictly outcome-based, meaning we only charge when the agent successfully resolves an inquiry or request. This model operates on a clear, deterministic framework where a billable resolution is defined by prioritizing the customer experience of resolution. Billing consumption is calculated based on the total number of these successful outcomes across your deployment channels.
Billable Resolutions
A Help Agent session qualifies as a Resolution when it satisfies all of these conditions.
- A session has at least two turns of messages between the user and agent (not abandoned)
- And the final feedback received from the user is not explicitly negative
- And the session either successfully concludes without the user escalating to a human, or the final feedback is explicitly positive
Any session that does not meet the conditions for a resolved session are considered unresolved and is not billed.
When a session is successfully resolved, any Agentforce actions or Data 360 queries run during that interaction are included and not billed separately.
| Billing Category | Description |
|---|---|
| Usage is calculated based on the number of interactions resolved by Help Agent through email. | |
| Enhanced Chat & Portal |
Usage is calculated based on the number of interactions resolved by Help Agent through Enhanced Chat or a portal. A resolution is defined as up to 2 hours of interaction. If the user interacts with the agent for more than 2 hours, it is counted as a second resolution. |
| Third-Party Messaging Channels | Usage is calculated based on the number of interactions resolved by Help Agent through third-party messaging channels such as WhatsApp, Facebook Messenger, or SMS. A resolution is defined as up to 2 hours of interaction. If the interaction exceeds 2 hours, it is counted as a second resolution. |
| Voice Channel | Usage is calculated based on the number of interactions resolved by Help Agent through a voice call. The duration limit for a single resolution is 10 minutes. Calls extending beyond 10 minutes are counted as multiple resolutions (for example, a 12-minute call is counted as 2 resolutions when resolved). |
Agentforce Voice Minutes Usage Type
The Agentforce Voice Minutes usage type is designed for Agentforce Voice Customers as an alternative to Voice Actions usage types. This Usage Type only applies to a limited subset of eligible customers who have purchased a subscription to Agentforce Voice Minutes. Contact your Account Executive for more details on Agentforce Voice Minutes.
| Billing Category | Description |
|---|---|
| Agentforce Voice Minutes |
Usage is determined by the duration of the voice call between an end user and Agentforce Voice. Usage is rounded up to the next full minute, meaning any fraction of a minute is treated as a full minute. For example,
If your org has both Agentforce Voice Actions and Agentforce Voice Minutes enabled, Voice Minutes take precedence for billing. When Agentforce Voice Minutes are available in your org, only Agentforce Voice Minutes are metered, and Voice Actions usage types are not applied to the same activity. |
Prompt Usage Types
Prompt usage types are calculated based on the number of calls to the large language model (LLM) gateway and vary depending on the large language model used. Prompt usage types apply to embedded generative AI features such as Service Replies that involve direct requests to the LLM for specific tasks. The specific category depends on the model that is used. See Large Language Model Support to find out which usage types apply.
| Billing Category | Description |
|---|---|
| Starter Prompts |
Usage is calculated based on two factors: the number of direct requests to the LLM via the LLM gateway, and whether the gateway uses a Bring Your Own Large Language Model (BYOLLM). Each starter prompt includes the processing of up to 2,000 tokens. Prompt usage is counted in chunks of 2,000 tokens, rounded up. Prompts that exceed this limit will be metered as multiple prompts, with each additional 2,000-token chunk counting as a new prompt. For example, a prompt with a total of 6,500 input and output tokens will be metered as 4 prompts. Tokens are units of data processed by the AI models. |
|
Standard Prompts Basic Prompts Advanced Prompts |
Usage is calculated based on two factors: the number of direct requests to the LLM via the LLM gateway, and whether the gateway uses a Salesforce managed large language model. The specific category depends on the model that is used. See Large Language Model Support to find out which usage types apply. All Standard, Basic, and Advanced prompts process up to 2,000 tokens per prompt. Token usage is rounded up in 2,000-token increments. All Standard, Basic, and Advanced prompts that exceed this limit will be metered as multiple prompts, with each additional 2,000-token chunk counting as a new prompt. For example, a prompt with a total of 6,500 input and output tokens will be metered as 4 prompts. Tokens are units of data processed by the AI models. Note Advanced Prompts were launched with a designated multiplier listed in the Agentforce
Flex Credits Rate Card. Salesforce will start metering the use of Advanced
Prompts on July 21, 2026.
|
Speech Foundation Usage Types
Speech Foundations are AI platform Speech-to-text, Text-to-speech and Translation services. These services empower admins and developers to integrate speech capabilities into workflows such as transcribing voicemails or generating audible alerts. Speech Foundation usage types are metered independently of any other usage types to ensure there are no overlaps in billing. Refer to your feature documentation to determine the applicable usage types for your feature.
| Billing Category | Description |
|---|---|
| Speech-to-Text | Speech-to-Text converts spoken audio into written text, a process run by automatic speech recognition (ASR) models. Usage is calculated based on the duration of the audio processed for each transcription, in seconds. In Digital Wallet, this usage is aggregated and displayed in minutes. |
| Text-to-Speech |
Text-to-Speech converts written text into natural-sounding spoken audio, a process run by automatic Speech Synthesis models. Usage is metered by the number of text characters processed for the speech generation. In the Digital Wallet, this usage is aggregated and displayed in units of million characters. To understand how your consumption appears in the Digital Wallet, consider a scenario where you process 9,000 characters for conversion from text to speech. Because this usage is in units of one million characters, your actual consumption is 0.009 units (9,000 divided by 1,000,000). In the Digital wallet, this usage is rounded to two decimal places and the value is displayed as 0.01 units. |
| Translation |
Translation converts text from one human language into text of another, a process executed by machine translation models. Translation usage is metered by the number of characters processed for the translation. To understand how your consumption appears in the Digital Wallet, consider a scenario where you process 9,000 characters for translation. Because this usage is in units of one million characters, your actual consumption is 0.009 units (9,000 divided by 1,000,000). In the Digital wallet, this usage is rounded to two decimal places and the value is displayed as 0.01 units. |
Buyer Engagement Agent Usage Types
Buyer Engagement Agent usage is measured by Leads Worked, which tracks when the agent engages a prospect through a supported channel. Outbound engagement counts when the agent interacts with a lead, while web conversations use one Leads Worked when the agent answers questions without attempting to create a lead and two when it also attempts to create a lead. Web usage is measured the same way whether anonymous visitor tracking is on or off. Engagements with the same prospect can count separately when they occur across different channels or identities. Related Data 360 services and agent actions can also consume Flex Credits, and you can use Digital Wallet to monitor usage across channels.
| Billing Category | Description |
|---|---|
| Leads Worked | Leads Worked measures how Agentforce engages prospects across channels, including web, email, SMS, and WhatsApp. Usage is deduplicated across all channels within a rolling 30-day window, so repeated interactions with the same prospect generally count as one Lead Worked. However, a web conversation can count as two Leads Worked when the agent also attempts to create a lead for the prospect. You can monitor usage in Digital Wallet to understand consumption and manage costs. To learn more about Leads Worked, see Leads Worked Billable Usage types. |
Other Flex Credits Usage Types
For information about Flex Credits usage types for Data 360, see Flex Credits Billable Usage Types for Data 360.