AI Chatbot Pricing Guide 2026: What Should You Pay?

AI chatbot pricing is one of the most confusing landscapes in SaaS. Every vendor uses a different model — per-agent, per-resolution, per-conversation, credits-based, flat-rate — and the sticker price on the website rarely reflects what you actually pay. Add-ons, overage fees, AI modules sold separately, and minimum commitments make apples-to-apples comparison nearly impossible without a spreadsheet.

This guide cuts through the complexity. We break down every major pricing model, show what the leading platforms actually cost at different team sizes and volumes, and give you a framework for calculating your real total cost of ownership — so you can make a decision based on numbers, not marketing pages.

The Four AI Chatbot Pricing Models

1. Per-Agent (Per-Seat) Pricing

The traditional SaaS model: you pay a monthly fee for each human agent who uses the platform. AI features may be included in the base plan or sold as an add-on. This is the model used by Zendesk, Freshdesk, Zoho Desk, and many traditional helpdesks.

How it works: base seat fee ($15–$165/agent/month depending on tier and vendor) plus optional AI add-ons ($25–$50/agent/month). Your bill scales with headcount, not ticket volume. A 10-agent team pays the same whether they handle 500 or 5,000 tickets.

The catch: per-agent pricing creates a perverse incentive. As your volume grows, you add agents, and your cost climbs linearly. Even if AI handles 60% of tickets, you still pay full seat price for agents who handle fewer conversations. And the AI capabilities that reduce agent workload often cost extra — you pay for the agents and the AI that is supposed to reduce the need for agents.

Best for: teams with stable headcount where agent collaboration features (internal notes, team assignments, workflows) matter more than automation depth. Worst for: fast-growing teams where volume outpaces headcount, making per-agent costs unpredictable.

2. Per-Resolution Pricing

You pay only when the AI successfully resolves a conversation without human involvement. This model is used by Intercom (Fin AI at $0.99/resolution) and increasingly by Zendesk (variable per-resolution fees on top of the AI add-on).

How it works: you pay a fixed fee for each AI-resolved conversation. If the AI escalates to a human, you do not pay the resolution fee (but you still pay the agent's seat cost). Pricing typically ranges from $0.50 to $2.00 per resolution.

The catch: per-resolution pricing sounds fair — you only pay for what works. But at volume, costs scale directly with success. A chatbot that resolves 3,000 conversations per month at $0.99 each costs $2,970/month just in resolution fees — on top of agent seat costs. The better your AI performs (higher resolution rate), the more you pay. This creates a perverse dynamic where the vendor profits most when your AI succeeds, but your costs also climb with success.

Best for: low-volume teams testing AI for the first time where the per-resolution cost is manageable. Worst for: high-volume teams where per-resolution fees at 2,000+ resolutions/month exceed what a flat or credits-based model would cost.

3. Credits-Based Pricing

You purchase AI usage credits that are consumed when the chatbot processes conversations. Credits cover AI processing (LLM calls, retrieval, generation), regardless of whether the conversation resolves or escalates. This model is used by Robylon AI and some newer AI-native platforms.

How it works: you buy a credits package (monthly or prepaid). Each AI conversation consumes credits based on complexity — a simple FAQ answer uses fewer credits than a multi-turn conversation with system integrations. There are no per-agent fees — you can add unlimited agents without increasing your platform cost.

The advantage: credits-based pricing decouples cost from headcount. Adding agents costs nothing. Your cost scales with AI usage, which correlates with ticket volume — but without the per-resolution penalty that charges you more when AI succeeds. And since there are no per-agent fees, growing teams get better unit economics as they scale.

Best for: growing teams where agent headcount fluctuates, high-volume operations where per-resolution fees would be expensive, and teams that want predictable AI costs without per-seat overhead. Worst for: very low-volume teams (under 100 conversations/month) where any subscription exceeds the value delivered.

4. Flat-Rate / Per-Ticket Pricing

Some platforms charge a flat monthly fee based on a ticket or conversation allotment. Gorgias uses per-ticket pricing ($10/month for 10 tickets up to $900/month for 5,000 tickets). Help Scout charges flat rates starting at $55/month with unlimited contacts.

How it works: you pay a fixed amount for a fixed allotment. Going over the allotment incurs overage charges or requires upgrading to the next tier.

The catch: per-ticket pricing punishes growth. A brand that grows from 1,000 to 3,000 tickets sees their Gorgias bill jump from $300 to $750/month — and the billing increments are not linear (you often pay for a full tier even if you are slightly over the previous one). Flat-rate models like Help Scout are more predictable but typically include limited AI capability.

Best for: very small teams with predictable, low volume. Worst for: growing brands where ticket volume is seasonal or increasing.

What the Major Platforms Actually Cost

Let us compare real costs for three team scenarios: a small team (5 agents, 1,000 tickets/month), a mid-size team (15 agents, 5,000 tickets/month), and a growth team (30 agents, 15,000 tickets/month).

Zendesk (Per-Agent + AI Add-On + Per-Resolution)

• Small team: Suite Growth ($89 × 5) + Advanced AI ($50 × 5) = $695/month + per-resolution fees (~$500–$1,000). Annual: ~$14,000–$20,000.
• Mid-size team: Suite Professional ($115 × 15) + AI ($50 × 15) = $2,475/month + per-resolution fees (~$2,000–$4,000). Annual: ~$54,000–$78,000.
• Growth team: Suite Professional ($115 × 30) + AI ($50 × 30) = $4,950/month + per-resolution fees (~$5,000–$10,000). Annual: ~$120,000–$180,000.

Intercom (Per-Seat + Per-Resolution)

• Small team: Essential ($29 × 5) = $145/month + Fin ($0.99 × 600 resolutions) = $594. Total: ~$739/month. Annual: ~$8,900.
• Mid-size team: Professional ($99 × 15) = $1,485/month + Fin ($0.99 × 3,000) = $2,970. Total: ~$4,455/month. Annual: ~$53,500.
• Growth team: Professional ($99 × 30) = $2,970/month + Fin ($0.99 × 9,000) = $8,910. Total: ~$11,880/month. Annual: ~$142,500.

Freshdesk (Per-Agent)

• Small team: Pro ($49 × 5) = $245/month. Annual: ~$2,940. (AI features limited at this tier.)
• Mid-size team: Pro ($49 × 15) = $735/month. Annual: ~$8,820.
• Growth team: Enterprise ($79 × 30) = $2,370/month. Annual: ~$28,440. (Freddy AI included but limited autonomous resolution.)

Gorgias (Per-Ticket)

• Small team (1,000 tickets): ~$300/month. Annual: ~$3,600.
• Mid-size (5,000 tickets): ~$900/month. Annual: ~$10,800.
• Growth (15,000 tickets): Custom pricing — typically $2,500–$4,000/month. Annual: ~$30,000–$48,000.

Robylon AI (Credits-Based)

• Small team: Free tier covers low volume. Paid credits for moderate usage: ~$100–$300/month. Annual: ~$1,200–$3,600.
• Mid-size team: Credits-based: ~$500–$1,200/month (no per-agent fees for all 15 agents). Annual: ~$6,000–$14,400.
• Growth team: Credits-based: ~$1,500–$3,000/month (no per-agent fees for all 30 agents). Annual: ~$18,000–$36,000.

How to Calculate Your True Total Cost of Ownership

The sticker price is never the full cost. Calculate your TCO by adding platform fees (base subscription, per-agent, per-resolution), AI add-on costs (if sold separately), overage and usage fees (credits, resolutions, tickets above your allotment), implementation costs (one-time setup, migration, configuration), integration development (custom API work for systems not natively supported), training costs (team onboarding and ongoing education), and maintenance (content updates, optimization time, admin overhead).

For a fair comparison, calculate the 12-month TCO for each platform at your current volume, then project at 2x volume to see how costs scale. The platform with the lowest year-1 cost is not always the best choice if it becomes the most expensive at 2x growth.

Pricing Red Flags

• AI sold as an add-on to every seat: Paying per-agent for AI that is supposed to reduce agent workload is double-dipping. Look for platforms where AI is a core feature, not an upsell.
• Per-resolution fees on top of subscription: Paying the platform fee AND paying per resolution means you are charged for the infrastructure and for each success. One or the other is fair — both is expensive.
• Opaque per-resolution pricing: If the vendor will not publish their per-resolution rate, it is likely high or variable. Ask for it in writing before signing.
• Minimum commitments with overage penalties: Annual contracts with low ticket allotments and high overage fees lock you into paying more if your business grows — exactly the wrong incentive.
• Features gated behind enterprise tiers: If basic AI, SSO, or reporting requires the most expensive plan, the entry-level pricing is misleading.

Bottom Line

The cheapest AI chatbot is not the one with the lowest sticker price — it is the one that delivers the highest resolution rate at the lowest total cost of ownership as you scale. Per-agent pricing penalizes growth. Per-resolution pricing penalizes AI success. Credits-based pricing (like Robylon) scales with usage without per-agent overhead, making it the most predictable model for growing teams. Calculate your 12-month TCO at current and projected volume, factor in all add-ons and hidden fees, and choose the model that gets cheaper per-ticket as your AI gets better — not more expensive.

AI pricing that rewards automation, not penalizes it. Robylon's credits-based model means no per-agent fees, no per-resolution surprises. Your cost scales with usage, not headcount. Start free at robylon.ai