A subscription chatbot costs from nothing to a few hundred dollars a month. A custom chatbot costs from a few thousand dollars for a simple rule-based bot to six figures for an AI bot connected to your systems, plus monthly running costs. The figures below are public and dated; they are what the market publishes, not our prices.
This guide is part of our technology guides. The difference between a chatbot and an agent that takes actions is in AI agents vs chatbots; the wider build economics are in how much custom software costs.
The three kinds of chatbot, and why the price differs
Almost every price confusion comes from comparing one kind with another. The kinds:
- Subscription platform bots. You configure a bot inside a help-desk or chat product and pay a monthly plan. Quick to start, limited to what the platform does, priced per seat or per conversation.
- Custom bots. Built for your site and your systems, often answering from your own documents with a language model, handing over to your people and writing to your CRM. Quoted per project, with running costs after launch.
- Enterprise systems. Bots across many channels, languages and business units, with security reviews, integration to core systems and a team to run them. Priced like any software program, in the hundreds of thousands.
Subscription chatbot prices, as published
Two platforms publish their prices plainly, so they make a fair reference. Tidio's pricing page (read 8 November 2026) lists a Free plan at $0, Starter at $24.17 a month with 100 billable conversations, Growth from $49.17 a month with 250, Plus from $300 a month, and a Premium plan on request; its Lyro AI agent is an add-on from $32.50 a month for 50 AI conversations. Intercom's pricing page (read the same day) lists seats at $29 (Essential, shown reduced to $19 for new customers at the time), $85 (Advanced) and $132 (Expert) per seat per month, with its Fin AI Agent from $0.99 per outcome and available without seats.
The pattern to note: the plan price buys the software and the human seats; the AI answers are metered separately, per conversation or per resolved outcome. A business with many conversations a month should budget the metered part first, because it grows with success.
Custom chatbot build costs: public ranges
Custom bots have no price list, so the public figures are guides' estimates and survey ranges. Elfsight's chatbot cost guide, with prices it states as verified as of February 2026, puts a simple rule-based custom bot at $5,000 to $30,000, an AI bot with natural language processing at $75,000 to $150,000, agency-built AI chatbots at $30,000 to $150,000, and enterprise systems at $200,000 to $1,000,000. It gives agency rates in North America of $100 to $200 an hour.
For the hourly side, GoodFirms' Custom Software Development Cost Survey 2026 (press release 17 March 2026, 100+ software companies) reports that 56% of the companies charge between $20 and $50 an hour, which is the band most remote development teams quote in. A custom bot of two to four weeks' work at those rates lands near the bottom of the guide ranges; one that connects to several systems and answers from a large document set lands near the top.
Running costs after the build
A custom AI bot keeps costing after launch. The four lines to budget:
- Model usage. An AI bot pays per token to the model provider. OpenAI's API pricing (standard tier, read 8 November 2026) ranges from $0.10 per million input tokens and $0.50 per million output tokens for its smallest current model to $10 and $50 for its largest, with mid-range models at $1 to $2 in and $5 to $10 out. A support bot answering a few thousand questions a month on a mid-range model usually spends tens of dollars, not thousands; the bill grows with volume and with how much document text is fed to each answer.
- Hosting and search. The bot, its document index and its logs need a server and a database. Elfsight's guide puts cloud hosting at $200 to $1,000 a month and language-model API fees at $50 to $300 a month for a typical custom build.
- Maintenance. Content changes, the model provider changes versions, the systems it connects to change. The same guide puts annual maintenance at 15 to 20% of the build cost.
- People. Someone reads the transcripts, fixes the wrong answers and handles the hand-overs. This is the cost most budgets leave out and the one that decides whether the bot helps.
What drives the price up or down
Six things move the quote:
- The job. Answering twenty fixed questions is cheap; answering anything in your help center needs a document index and a model; booking, quoting or changing an order needs integrations and the care of an agent, as our guide to AI agents vs chatbots explains.
- Connections. Each system the bot reads from or writes to (CRM, order system, calendar, help desk) adds build and maintenance work.
- Content. A bot is only as good as the answers it is given. Writing or cleaning the source content is often a third of the project.
- Hand-over. Routing to a person, with the conversation attached, on the channels your team uses.
- Languages and compliance. Each language is tested separately; privacy rules in your market decide what the bot may store and where.
- Who builds it. Hourly rates differ several times over between regions, as the GoodFirms and Elfsight figures show; the total depends more on the scope than on the rate.
Which kind should you buy?
Start with a subscription bot if your questions are mostly the same twenty, your team already uses the platform, and you want to learn what customers ask before spending more. Move to a custom bot when the answers live in your own documents and systems, when the bot must write to your CRM or order system, or when the platform's per-conversation metering costs more than a build would. Treat an enterprise system as a program with a team, not a purchase. The table below sets the three side by side.
How to budget for a chatbot
Six steps:
- Write the job in one paragraph. What it answers, what it does, who it hands to, where it runs.
- Count the conversations. Last year's chats, emails and calls on the same topics; this sizes the metered costs.
- Choose the band. Subscription, custom or enterprise, from the job and the count.
- Budget the running costs for a year, not only the build: model, hosting, maintenance and the person who reads the transcripts.
- Get two or three scoped quotes for a custom build, each itemised against the same paragraph, and compare the scope before the price.
- Hold back for content and testing. The bot that launches with checked answers and a hand-over path is the one customers keep using.
Conclusion
A chatbot costs what its kind costs: a published monthly plan for a subscription bot, a scoped project price plus running costs for a custom bot, a program budget for an enterprise system. The published prices and public ranges above, all dated, give the bands; the scope decides where inside a band your bot lands. Budget the running costs and the people, not only the software.
PromotEdge International is an international strategy, technology, marketing and design company, and our delivery team is in India. Our chatbot development and AI agents teams scope the job first and price against it, build bots that answer from your own content and hand over to a person, and run them within the budget agreed.
Facts we rely on: see our company facts for what we state about ourselves and where each fact comes from.