Chatbot pricing, and what an AI chatbot costs

Chatbot pricing for a small website runs from nothing on a free plan to a few hundred dollars a month. We priced the same month of 1,000 questions on eleven tools, and the bills ran from $12 to $227.10. Most of that gap comes from what each vendor counts as one unit, a seat, a credit, a resolution or a month, and from whether that count grows with your traffic. Every price here is for monthly billing, read on the vendor’s own page on 23 September 2026.

Deacon is an AI customer support agent for founders who answer support themselves. It charges a set price for a month of answers, so a busy week never adds to the bill, and its free plan gives you 50 answers a month with no card. It answers your customers in seconds from your own docs, and shows you what they’re stuck on and what they want next.

The four ways vendors charge

AI chatbot pricing is built from one of four units, and some bills stack two of them.

• By the seat. You pay for each person on your team, and the AI comes with an allowance or a meter of its own.
• By the message or credit. You buy a pool of replies, and a bigger model spends the pool faster.
• By the resolution or conversation. You pay for each conversation the AI settles, or for each one it joins.
• By the month. One price buys an allowance, whatever model writes the replies.

The month we priced

The site has one founder and takes 1,000 questions a month from visitors. The bot writes one reply to each question, so that’s 1,000 replies.

Questions arrive in conversations, and the per-conversation bills depend on how many. Gleap’s pricing page says a conversation with its AI runs about 4.3 messages. Deacon measured its own on 22 September 2026 at 1.95 answers across 1,053 conversations. We use 4 replies a conversation, which makes 250 conversations and keeps the per-conversation bills low. At 2, there are twice as many conversations, and those bills rise to what the 2,000-question column further down shows.

Not every conversation ends with the problem solved. Intercom says Fin’s average resolution rate is 76% across more than 8,000 customers, Help Scout puts its own at 73%, and HubSpot’s pricing page claims 70% or more. We use 76% for every sum that bills by the resolution, which is 190 resolved conversations.

So the month is 1,000 questions and 1,000 replies, in 250 conversations, 190 of them resolved by the AI, with one person on one seat. Where a vendor lets you choose the model, every sum uses its lightest one.

Priced by the seat

“You pay per user on your team, and each plan comes with a set number of AI Agent resolutions per month,” says ChatBot.com’s pricing page. It counts one resolution when the AI solves a customer’s question, however many questions were in the chat, so our month needs 190.

Essential is $25 a user and includes 10 resolutions. Past those, ChatBot.com adds a pack of 50 resolutions for $49.50 whenever you run out, which works out at $0.99 each. Our other 180 take 4 packs, or $198, so the month on Essential comes to $223. Growth is $99 a user and includes 200 resolutions, enough on its own, so the per-seat month is $99.

Front and Zendesk start from a seat as well and charge for the AI on top. Front bills $0.89 for each conversation its AI resolves, and $0.39 for one that its workflow hands to your team. Zendesk’s pricing page includes 5 to 15 automated resolutions per agent a month, by plan, and charges $2.00 for each one past them on pay as you go.

Priced by the message or credit

These vendors sell a monthly pool. SiteGPT counts a visitor’s question and the bot’s reply as a message each, so our month is 2,000 of them, as SiteGPT pricing explains. Dante AI says “a credit is one answer from your AI agent”, and its lightest models spend one credit a reply. Chatling charges each reply “a certain number of AI credits based on the model used”, half a credit on its lightest models.

Each Starter plan covers the month with room to spare, as long as a small model writes the replies. A bigger model drains the pool faster, as Chatling pricing shows model by model. Dante AI’s cost article puts a mid-tier model at five credits a reply, which makes our month 5,000 credits. Starter holds 2,000, and the other 3,000 come from top-ups of 1,000 for $20 each, so the month becomes $100.

Chatbase prices its credits the same way, at 1 to 6 a reply by model. Its Hobby plan, at $40 a month, holds 700, so our month needs 300 more, and its auto recharge sells them at $40 for 1,000 that don’t expire. There’s more in Chatbase pricing.

Priced by the resolution or the conversation

Intercom charges $0.99 for each Fin outcome, and at most one a conversation. Nothing is charged when Fin can’t help or the customer asks for a person. Intercom’s help centre adds that a customer who leaves after Fin’s answer without asking for more help “is counted as an Assumed Resolution and will be charged.”

Our 190 resolutions come to $188.10. On Intercom’s own helpdesk, Fin needs at least one seat, and Intercom is testing two prices for Essential on the same page, $29 or $39 a seat on monthly billing. So the month is $217.10 or $227.10, depending on which price you’re shown. Fin can also run on a helpdesk you already pay for, with no seat and a minimum of 50 outcomes a month.

Help Scout’s AI Answers costs $0.75 a resolution, and a conversation doesn’t count if the customer asks for more help. Our 190 come to $142.50, and a Standard seat at $30 makes the month $172.50. HubSpot’s AI customer agent charges $0.50 a resolved conversation, but only on Service Hub Professional and Enterprise, and Professional comes with an annual commitment and a one-time onboarding fee of $1,500.

Tidio’s Lyro charges for the conversation instead, resolved or not. In Tidio’s words, a Lyro conversation is one “that has at least one reply from our AI Agent”, so all 250 of ours count. Lyro is sold in fixed packages, and the smallest that holds 250 is 300 conversations for $210 a month, however quiet the month turns out. Tidio’s help article on the Lyro limit says a thread closes after 15 idle minutes and a new message starts another, although its pricing page says a visitor who comes back is still one conversation. There’s more in Tidio pricing.

Priced by the month

“Flat pricing, AI included,” says Chattering’s pricing page. A flat plan charges one price for the month and one allowance, whatever model writes the replies, and past the allowance you move up a plan.

Chattering needs its Pro plan here, because Starter holds 500 replies, half our month. When the replies run out, its AI stops for the rest of the month and asks visitors for their email.

Deacon’s Starter holds exactly 1,000 answers, and our month uses every one. Follow-up questions count, and so does a reply saying it doesn’t know, while chats in the Playground never do. A slightly busier month means Growth at $150, or for the rest of that month, a short reply that asks for the visitor’s email unless they’ve already left it.

Crisp calls its price “a flat pricing per workspace”, and for seats it is. The AI runs on a pool of credits inside each plan, $5 on Mini and $25 on Essentials, which Crisp reckons at around 90 and 450 conversations. Crisp’s help centre puts an average conversation at $0.05 to $0.10, so our 250 need Essentials, at $95. When the pool runs dry, Hugo hands new conversations to your team unless you switch on pay as you go, at a rate Crisp doesn’t publish. There’s more in Crisp chat pricing.

The same month, every way

At 2,000 questions there are 500 conversations and 380 resolutions. ChatBot.com’s Growth covers 200 of them, and the other 180 take 4 more packs, $198. Fin is 380 outcomes at $0.99 plus the seat, and Help Scout 380 resolutions at $0.75 plus the seat. Tidio moves to its 500-conversation package, as long as you switch off its automatic upgrade, which is on by default for a new project on a monthly plan and moves a package up a level once 95% of it is used. Crisp reckons Essentials at around 450 conversations, so 500 means Plus, or pay as you go at the rate it doesn’t publish.

Doubling the traffic adds $200 to Crisp, $198 to ChatBot.com, $188.10 to Fin, $142.50 to Help Scout, $140 to Lyro, $100 to Deacon and $12 to Elfsight. It adds nothing to Chattering, or to the credit and message plans on their lightest models.

None of this says whether the answers are any good, and a cheap bot that makes things up costs you more than it saves. We’ve written about how to tell the difference separately.

Intercom, Help Scout, Tidio, Crisp, ChatBot.com and Chattering all come with an inbox for your team to work in. If two or three people will reply from one every day, part of those bills buys something you’d pay for anyway.

What changes the numbers for your site

The number of questions in each conversation matters more than anything else to the vendors that bill by the resolution or the conversation. At 2 questions a conversation instead of 4, the same 1,000 questions make 500 conversations and 380 resolutions, and those bills rise exactly as they do when traffic doubles. The message, credit and flat plans stay put, because the number of replies hasn’t changed.

Each vendor’s small print moves the count as well. Tidio’s 15-minute rule, Intercom’s assumed resolutions and SiteGPT’s question-and-reply pairs all change what the same traffic costs. At the limit, ChatBot.com adds its packs by itself, and on its Plus plan Botpress adds packs of 100 conversations for $65 once you reach 95% of your quota, which its pricing FAQ says you can’t turn off. We’ve worked out what that does to a month in Botpress pricing.

You can price your own site from last month’s inbox or chat history. Count the questions people asked, which is also your count of replies. Divide that by the number of conversations they came in, and you have your own questions per conversation in place of our 4. The conversations nobody on your team had to finish are roughly what a per-resolution vendor would bill.

What Deacon costs, and what the price covers

Deacon bills a set price for a month of answers. The free plan gives you 50 answers a month, with no card and no end date. Starter is $50 a month, or $40 a month billed yearly, for 1,000 answers. Growth is $150 a month, or $120 a month billed yearly, for 4,000. A conversation takes about 2 answers, so that’s about 25, 500 and 2,000 conversations.

Every feature is on every plan, the free one included, and the plans differ only in answers, knowledge, pages from your website and seats. Nothing is billed past your plan’s answers, and no plan upgrades itself. Your team’s owners get an email when 80% of a month’s answers are used, and another if they run out. Every plan answers with the same model, so no credits run out faster on a better one, and there’s no badge to pay to remove.

For that price, Deacon answers your customers in seconds from your own docs, in the language they write in. When your docs don’t cover a question, it says it doesn’t know instead of guessing, and asks for the visitor’s email. If they leave it or ask for a person, you get an email, and if they still have the chat open, your reply from the dashboard appears there. Messages between a visitor and you never count as answers.

Deacon also gets better the more you answer. Every question your docs didn’t cover waits in your dashboard with an Answer this button, and what you write becomes part of what Deacon knows, so it can answer the next person who asks. Topics groups what people ask and labels most of it as a how-to, a bug report, a feature ask and so on, so you know what to fix and what to build next.

Questions about what an AI chatbot costs

How much does an AI chatbot cost for a small business?

For one person answering support, a hosted AI chatbot runs from nothing on a free plan to a few hundred dollars a month. In our month of 1,000 questions, the bills ran from $12 to $227.10. The seat price is rarely what decides it. The meter is. The best AI chatbots for customer service lists nine with how each bills its AI, and AI chatbot for small business covers which questions one can take for a shop, clinic, agency or trade.

Is there a free AI chatbot?

Yes, several, though each caps its AI replies, and some never renew. We compared them, and what each one’s visitors see when it runs out, in free AI chatbots for a website.

How much does it cost to build your own AI chatbot?

Building your own swaps the subscription for the model’s usage fees, paid by the token, plus the time to build and run everything around it. That means finding the right passages in your docs for each question, keeping every conversation and handing chats to a person. Unless chat is what you sell, a hosted chatbot is cheaper once your time is counted.

What happens when Deacon’s answers run out?

Visitors get a short reply saying Deacon can’t look into that one right now, and are asked for their email unless they’ve already left it. You can still read and reply to every conversation, nothing is billed past your plan’s answers and your plan never upgrades itself.