Independent reference. aiagentforcustomerservice.com is not affiliated with, endorsed by, or sponsored by Intercom, Zendesk, Salesforce, HubSpot, Decagon, Sierra, Forethought, Ada, Cognigy, Kustomer, Freshworks, Tidio, or any AI agent vendor named on this site. Pricing compiled from public sources. Verify with vendor before purchasing. Last verified April 2026.

AI

AI Customer Service Pricing Models in 2026: Per-Resolution vs Per-Conversation vs Per-Seat vs Flex Credits

Four distinct pricing models are in active use across the 12 vendors compared on this site. Which costs less at your volume depends on your deflection rate, agent headcount, and helpdesk. Below: definitions, vendor examples, and worked math at 5K and 25K conversations/month.Last verified April 2026.

The 4 Pricing Models

Per-resolution

Pay only when AI fully resolves without human escalation.

DEFINITION: Zendesk: no human escalation in resolution window. Intercom Fin: end of conversation without escalation. HubSpot Breeze: resolved conversation with no negative feedback. Sierra: pre-negotiated outcome.

Per-conversation

Pay for every AI interaction whether resolved or not.

DEFINITION: Any conversation where the AI responds, regardless of resolution. Salesforce counts each unique customer-facing conversation. Kustomer counts each engaged conversation (AI sends at least one response).

Per-seat

Flat per-agent monthly fee regardless of conversation volume.

DEFINITION: Price is fixed per agent per month. Volume does not change the unit cost. Best when volume is predictable and high relative to headcount.

Consumption credits

Buy credits in packs; AI actions consume credits at varying rates.

DEFINITION: Credits are bought upfront and consumed by AI actions. Simple actions (FAQ response) use fewer credits than complex actions (multi-step workflow execution). Best for variable workloads where per-conversation pricing would be volatile.

Worked Examples: Cost at Real Volume

All calculations use the vendor's headline rate. Implementation costs and mandatory seat minimums are noted but not added to the monthly figure.

5,000 conversations/month
40% deflection = 2,000 resolutions
VendorModelMonthly AI CostAnnual AI CostWhat's not included
HubSpot BreezePer-resolution$1,000$12,000+ Service Hub $50+/seat + $3K-$7K onboarding
Freshdesk Freddy (5 agents)Per-seat$390$4,6805 agents x ($49+$29 Copilot) + 500 free sessions
Intercom FinPer-resolution$1,980$23,7602,000 resolutions x $0.99
Zendesk AIPer-resolution$2,600$31,2002,000 x $1.30 (101-1,000 tier)
Kustomer Bundle (8 seats)Per-seat bundled$1,032$12,3848 x $129 (AI included) - minimum 8 seats
Salesforce AgentforcePer-conversation$10,000$120,0005,000 x $2 (per conversation, not per resolution)
25,000 conversations/month
50% deflection = 12,500 resolutions
VendorModelMonthly AI CostAnnual AI CostWhat's not included
HubSpot BreezePer-resolution$6,250$75,00012,500 x $0.50 + Service Hub seats
Zendesk AI (tiered)Per-resolution$12,500$150,00012,500 x $1.00 (5K+ tier)
Intercom FinPer-resolution$12,375$148,50012,500 x $0.99
Freshdesk Freddy (15 agents)Per-seat$1,170$14,04015 x $78 (Pro + Copilot) + session overages
Salesforce Flex CreditsConsumption~$5,000-$8,000~$60K-$96KDepends on action complexity and Foundation credits
Decagon (enterprise)Per-conversation~$33K-$67K~$400K median ACVEnterprise quote only; Vendr marketplace data

Decision Framework: Which Model at Which Volume?

Sub-2K conversations/month
Per-seat (Freshdesk Freddy, Tidio)
At this volume, per-resolution pricing starts below $200/mo but flat-billing tiers are similar and simpler to budget.
2K-25K conversations/month
Per-resolution (HubSpot Breeze $0.50, Fin $0.99)
The per-resolution model's alignment with outcomes makes sense here. HubSpot Breeze is cheapest at $0.50 for HubSpot-native shops.
25K-100K conversations/month
Zendesk tiered ($1.00 at 5K+ tier) or Salesforce Flex Credits
Zendesk's tiered rate ($1.00 at 5K+ resolutions) beats Intercom Fin's flat $0.99 and is cheaper than Salesforce's $2/conversation for Zendesk shops.
100K+ conversations/month
Enterprise quote-based (Decagon, Sierra, Forethought)
At this scale, enterprise-negotiated rates per-conversation often compete with public per-resolution pricing on total cost, with more customisation.

Common Questions

What is per-resolution pricing for AI customer service?

Per-resolution pricing means the vendor charges only when the AI fully resolves a customer issue without human escalation. Vendors: Intercom Fin ($0.99), Zendesk AI ($1.00-$2.00 tiered), HubSpot Breeze ($0.50, updated April 2026), Sierra (outcome-based custom). The buyer only pays for successful outcomes. The risk is that each vendor defines 'resolution' differently.

How does per-resolution pricing differ from per-conversation pricing?

Per-resolution charges only for successful AI resolutions (no human escalation). Per-conversation charges for every AI interaction whether resolved or not. Salesforce Agentforce ($2/conversation) and Kustomer AI for Customers ($0.60/conversation) use per-conversation. At a 40% deflection rate, per-resolution at $0.99 costs the same as per-conversation at $0.40. Above 40% deflection, per-resolution is more expensive per ticket processed but you only pay for successes.

When does per-seat pricing beat per-resolution?

Per-seat pricing (Freshdesk Freddy $29/agent, Kustomer $89-$139/user) beats per-resolution when your volume is high relative to your headcount. At 3,000+ conversations per agent per month, a $29/seat flat fee is cheaper than $0.99/resolution at 40% deflection (1,200 resolutions x $0.99 = $1,188 vs $29). The crossover point for Freshdesk Freddy is approximately 30 resolutions per agent per month.

What is the cheapest AI customer service pricing model?

At low volumes (sub-2K conversations/month), per-seat is usually cheapest. At 2K-25K resolutions/month, per-resolution wins (HubSpot Breeze $0.50 is the lowest rate). At 25K-100K, Zendesk's tiered pricing ($1.00 at 5K+ resolutions) or Salesforce Flex Credits become competitive. Above 100K, enterprise quote-based vendors (Decagon, Sierra) can be cheaper per-resolution than public pricing at scale.