An AI chatbot costs anywhere from $0 to $150,000 or more in 2026, and the range is that wide because the word "chatbot" now covers everything from a free no-code AI agent you launch in under a minute to a bespoke enterprise system built by a software team. A no-code AI platform typically runs $0 to $400 per month. A custom development project typically runs $10,000 to $150,000 or more upfront. An agency-managed build usually lands between the two, with a monthly retainer on top. Most small and mid-sized businesses today get a production-quality AI agent live for under $100 per month.
This guide breaks down every path by real price, explains the pricing models you will be quoted, shows how the number changes as you scale, and lists the costs almost nobody puts on a pricing page. If you want a fast answer for your own situation, the five-step estimate further down walks you through it.
Chatbot cost at a glance
Three build paths dominate the market, and they price in fundamentally different ways. A platform charges a predictable monthly subscription. A custom build charges a large one-off fee plus ongoing running costs. An agency charges a setup fee plus a recurring retainer. Here is how they compare on the numbers that decide the budget.
| AI platform (no-code) | Custom development | Agency build | |
|---|---|---|---|
| Setup cost | $0 | $10,000 to $150,000+ | $2,000 to $25,000 |
| Monthly cost | $0 to $400 | $500 to $5,000 (hosting, model usage, maintenance) | $500 to $5,000 retainer |
| Typical first-year total | $0 to $4,800 | $16,000 to $200,000+ | $8,000 to $85,000 |
| Time to live | Minutes to hours | 3 to 9 months | 2 to 8 weeks |
| Maintenance | Included | Your engineering team | Included in retainer |
| Best for | Self-serve businesses of any size | Deeply bespoke workflows at enterprise scale | Teams that want hands-off setup |
For most readers the honest headline is the first column: the platform path has quietly absorbed the majority of use cases that used to justify a $20,000 project. The rest of this guide explains when that is true for you, and when it is not.
What actually drives the price of a chatbot
Two chatbots can differ in price by a factor of a thousand, and the gap comes down to a handful of variables. Before you compare quotes, know which of these apply to you, because a vendor's number only makes sense once you know what it includes.
- Build path. Buying configurable software is cheap. Paying engineers to build software is expensive. This is the single biggest lever on the whole number.
- Complexity of the job. Answering questions from your website is standard and inexpensive. Custom conversation logic, transactions, and unusual compliance rules add cost fast.
- Conversation volume. Every AI answer has an underlying model cost. Higher volume means a higher plan tier or a larger model bill, so your traffic is a direct input to the price.
- Integrations. Connecting to your help desk, CRM, or booking system is straightforward on a platform and a line item on a custom build.
- Languages and channels. Multilingual support and multiple channels can be included on a platform, or extra engineering elsewhere.
- Who maintains it. A subscription bundles upkeep. A custom build makes upkeep your permanent responsibility. See how AI chatbots work for what sits under the hood of each answer.
Path 1: AI chatbot platforms ($0 to $400 per month)
Modern AI chatbot platforms train on your existing content, your website, your documents, and your knowledge base, then answer customer questions with no code and no project plan. This category is what people mean today when they search for a conversational AI platform, and it has effectively replaced the custom build for the majority of use cases.
Real pricing, using Dante AI's 2026 plans as a concrete example:
- Free: $0. Build, train, and test an AI agent on your own website content without paying anything. New accounts get up to 700 free credits during onboarding and 100 free credits every month after that, so you can watch it answer real questions before you spend a dollar. This is the whole point of a free AI chatbot tier: prove the value first.
- Starter: $40 per month. Production usage for a small business: 2,000 monthly credits, access to advanced AI models, and integrations with the tools you already use.
- Advanced: $120 per month. 8,000 monthly credits for businesses whose AI agent handles a meaningful share of their customer conversations.
- Pro: $400 per month. 30,000 monthly credits for high-volume support operations, the tier where an AI agent is doing the work of a support team.
Usage beyond a plan is typically metered in small top-ups, for example $20 per 1,000 credits, rather than surprise overage bills, and you can move between tiers as volume changes. Because the software is already built, a no-code chatbot platform gives you the roughly 95 percent of functionality almost every business needs. The honest limitation: if your requirements are genuinely bespoke, deep integration with proprietary internal systems, unusual compliance regimes, or custom conversation logic that no configuration surface covers, you are in custom-build territory.
Path 2: Custom development ($10,000 to $150,000+)
A custom chatbot is a software project, so you are paying for engineers, project management, infrastructure, and ongoing ownership rather than for a subscription. This is the path behind searches for how much it costs to build or develop a chatbot, and the commonly quoted 2026 ranges are:
- Simple rules-based flow: $10,000 to $30,000. Scripted paths and limited understanding, the kind of scripted experience that made early chatbots unpopular.
- AI assistant with retrieval: $30,000 to $80,000. A model connected to your knowledge with custom logic, which is exactly what platforms now deliver off the shelf.
- Enterprise conversational system: $80,000 to $150,000 or more. Multi-system integration, custom compliance, and dedicated infrastructure.
The number on the proposal is never the whole number. Custom builds carry ongoing costs that commonly add 20 to 40 percent of the build price every year: model API usage, hosting, monitoring, upgrades, and the engineering time to keep it working as your content and systems change. Budget for the running cost with the same seriousness as the build cost, because over three years the running cost often exceeds it.
Path 3: Agency builds ($2,000 to $25,000 setup, plus retainer)
Agencies typically configure a platform or assemble open-source components for you, then charge a monthly retainer to manage it. Searches for chatbot agency pricing and chatbot-as-a-service pricing land here. This path makes sense when you have no internal capacity at all and want a single accountable partner to own the whole thing.
The trade-off is speed and dependency: changes route through the agency, and the retainer runs whether or not anything changed that month. A common pattern is to have an agency stand the AI agent up on a platform, then bring management in-house once your team is comfortable, which converts a recurring retainer into a much smaller platform subscription.
Chatbot pricing models explained
Vendors price the same product in different ways, and comparing quotes is only possible once you can name the model. These are the five you will actually be quoted:
- Flat monthly subscription. One predictable price for a bundle of usage. Easiest to budget, and the norm for small and mid-sized plans.
- Metered credits or messages. You pay for what you use, often as credits where heavier AI models cost more per answer. This keeps cost aligned with real activity and is how most platforms handle overage.
- Per-conversation or per-resolution. You pay each time the AI agent handles a conversation, or only when it fully resolves one. Attractive at low volume, but it can get expensive fast as traffic grows.
- Per-seat. Priced by the number of human agents who use the tool. Common on legacy support suites, and worth watching because it charges for people rather than for automation.
- One-off build fee. The custom-development model: a large upfront project cost, then running costs you own. No subscription, but no included maintenance either.
Most no-code platforms blend the first two, a flat subscription with metered credits, so your bill tracks usage without nasty surprises. When a quote looks unusually cheap or expensive, the pricing model is usually why, so always ask which one you are being sold.
How chatbot cost changes with business size
The same question, "how much does a chatbot cost", has a different answer at each scale, because volume and complexity both climb as you grow. Here is the realistic 2026 spend by business size.
- Small business: $0 to $120 per month. A free tier to start, then a Starter or Advanced plan once the AI agent is proven. Live within a day, no engineering required.
- Mid-market: $120 to $400 per month. Higher conversation volume, more integrations, and multilingual support, all still comfortably inside platform territory.
- Enterprise: $400 to a few thousand per month on a platform, or a $50,000-plus custom build when a hard requirement genuinely cannot be met off the shelf. If you are weighing a large deployment against staffing, compare it directly with the cost of implementing AI in a call center.
The pattern to notice: for the vast majority of businesses, growing does not mean leaving the platform path, it means moving up a tier on it. The jump to a custom build is driven by unusual requirements, not by size alone.
The hidden costs nobody quotes upfront
Whichever path you choose, these are where chatbot projects actually get expensive, and where a low headline price can hide a high real one.
- Content and training data. An AI agent is only as good as what it learns from, so preparing clean, current source content is a real cost. Outdated help docs produce a confidently wrong AI agent. Platforms that train directly on your live website drive this close to zero; custom builds often burn weeks here, which is why "chatbot training data pricing" is a question people ask.
- Model usage. Every AI answer has an underlying model cost. Platforms bundle it into plan credits. Custom builds pay the API bill directly, and it scales with traffic.
- Maintenance and retraining. Your product changes, your policies change, and the AI agent must follow. This is included in platform subscriptions and a permanent line item everywhere else, commonly 20 to 40 percent of a custom build's cost each year.
- Human handover. The moment a conversation needs a person, your AI agent needs somewhere to send it. Check whether handover is standard, a paid add-on, or a custom integration.
- The cost of a wrong answer. The most expensive chatbot is the one that answers incorrectly and drives a customer away. Whatever you build, test it against real customer questions before it faces customers.
How to estimate your chatbot cost in five steps
You do not need a quote to get a realistic number. Work through these five steps and you will land inside the right range in a few minutes.
- Define the job you need done. Write down exactly what the AI agent must do. The narrower and more standard the job, the cheaper the path.
- Estimate your monthly conversation volume. Count the customer messages you handle in a typical month. Volume is the biggest driver of ongoing cost.
- Match volume and complexity to a build path. A standard job at normal volume points to a platform at $0 to $400 per month. Genuinely bespoke logic points to a custom build at $10,000 or more.
- Add content prep, integrations, and maintenance. Fold in the hidden costs above. Platforms include most of them; custom builds price each one separately.
- Validate on a free tier before you pay. Build it free, train it on your real site, and test it on the questions your customers actually ask.
Is a chatbot worth the cost?
A well-trained AI agent regularly resolves the majority of routine questions without a person, around the clock, in multiple languages. Measured against the cost of one support hire, even the Pro tier is a fraction of the price, and the free tier costs nothing to trial. That is why the worth-it question is really an accuracy question.
The economics only work when the answers are good, so the decision is not "should I pay for a chatbot" but "can this AI agent answer my customers correctly". Prove that first on a free plan, and the cost question mostly answers itself.
What should you actually pay in 2026?
- Testing the idea: $0. Use a free tier, train it on your site, and judge the answers yourself. If a chatbot cannot prove its value for free, do not pay for it.
- Typical small business: $40 to $120 per month, live within a day.
- High-volume support operation: $400 to a few thousand per month depending on volume, still a fraction of one support hire.
- Genuinely bespoke enterprise needs: $50,000 or more and months of build time, justified only when a platform demonstrably cannot meet a hard requirement.
Want the real answer for your business? Train an AI agent on your website in a few minutes and see how it handles your actual customer questions. Start free and judge the answers before you spend anything.
Further reading
Keep going with these guides from the Dante AI library: