Agentforce Flex Credits Explained

With the arrival of Agentforce, the game has changed. We are moving away from simple chatbots and toward autonomous agents that can actually do work. But how do we pay for it? Historically, Salesforce has used various pricing models per user, per license, per conversation or data-based consumption. For Agentforce, Salesforce has introduced a more granular model: Flex Credits.

Here is everything you need to know about Flex Credits and how they will impact your Agentforce strategy.

How Flex Credits Are Consumed

In the past, Agentforce was priced at a flat rate of $2 per conversation. While simple, this didn’t always account for the work the Agent needs to do. You’d pay the same $2 for a visitor/ prospect asking for your store hours as you would for an agent troubleshooting a complex technical issue using multiple data sources.

Flex Credits are Salesforce’s answer to this imbalance. Instead of paying for a conversation, you pay for actions. This shifts the model from a flat fee to value-based pricing. You pay more for complex tasks and less for simple ones.

The most important thing to understand about Flex Credits is that they are metered by Actions. An action is a specific function your agent executes, like looking up a record in Data Cloud, updating a lead status in Sales Cloud, or searching for knowledge articles.

Here is the breakdown of how consumption works:

1. Per-Action Cost

Each action taken by an agent typically costs $0.10. Because a single interaction or “conversation” might involve several steps, the total cost is the sum of those steps. Simple Interaction: An AI agent answers a basic question using one knowledge base lookup. This might be 1 or 2 actions ($0.10–$0.20). Complex Interaction: An AI SDR (Sales Development Representative) agent researches a lead, checks their recent website activity via Data Cloud, summarises the findings, and then books a meeting in a rep’s calendar. This example may take up to 6 actions ($0.60).

2. Only Pay for What Works

One of the biggest benefits of the Flex Credit model is that if the agent doesn’t take an action, you aren’t charged. If a user says “Hello” or “Thanks!” and the agent responds without querying a database or executing a flow, no credits are consumed. You are only billed when the agent actually “does work” for you.

3. Granularity = Better ROI breakdown

Flex Credits allow you to see exactly which types of AI interactions are most expensive and which are most efficient, making it much easier to justify the spend compared to a flat $2 per conversation fee. While Flex Credits are the flagship for usage-based pricing, Salesforce still offers an unmetered per-user, per-month license for specific roles in Sales and Service. Choose Flex Credits if you have high volume, variable interactions where many tasks are simple, and you want to pay only for exact “work” performed.

Choose Unmetered Pricing if you have power users (like full-time SDRs) who will use Agentforce constantly and want a predictable monthly bill.

Top Tip: If you’re just starting, use the Digital Wallet in your Salesforce org. It provides near real-time usage data so you can see exactly how your agents are spending your credits before you commit to more.

Consumption Pricing Breakdown

Salesforce offers different buying models. The table below summarises those options, their current cost and why they might benefit you.

For most businesses, the Flex Credit model offers the most transparency – you aren’t paying for empty seats, but for executed tasks.

For high-volume support teams, compare the cost of “actions” vs flat “conversation” rates to see which fields perform better ROI – Salesforce provides a handy calculator for this exact scenario.