AI Chatbot Cost in 2026: $0 to $999/Month

Quickchat AI’s Free plan is for personal, non-commercial use and includes 50 monthly AI credits. Commercial use starts at $9/month, with self-service plans up to $999/month, excluding applicable VAT. Your budget depends on AI replies, the selected model’s credits per reply, testing and required features. Website visits and conversations alone do not determine the price.

The examples below work through a small shop, an existing helpdesk and higher-volume support. Each separates the platform allowance from human work, implementation and external services. Enterprise pricing starts at $0.50 per resolution with a minimum of 2,000 resolutions per month; a custom implementation needs its own scoped quote.

The free-plan walkthrough shows what spends credits in a fictional, non-commercial demo.

Optional setup tools

• Compare build and subscription costs. Calculate costs from your supplied implementation, maintenance and subscription assumptions. This does not estimate missing rates or recommend a purchase.

• Compare supplied pricing quotes. Normalize declared pricing assumptions and flag missing inputs. This does not fetch vendor prices or supply a commercial quote.

• Check plan requirements. Compare declared usage, content, team and feature needs with public Quickchat AI plan bundles. This is not a price quote or account check.

• Compare replies per credit budget. Calculate the maximum whole AI replies for each supported model from a supplied credit budget. This does not check your balance or model eligibility.

Chatbot cost at a glance

Build a budget from the following inputs. Prices and billing terms in this guide were checked on November 5, 2026; confirm the applicable plan and contract before buying.

In this guide:

• How much does an AI chatbot cost?
• What makes a chatbot more expensive?
• Chatbot pricing models explained
• Is it cheaper to build or buy a chatbot?
• We built an AI chatbot on the free plan
• Hidden chatbot costs people miss
• How to estimate your monthly chatbot cost

How much does an AI chatbot cost?

These are illustrative monthly workloads, not customer averages or quotes. Each assumes one credit per AI reply, including testing. A conversation means an actual chat with the agent; it is not a website visit or a proven resolution.

Prices use monthly billing and exclude applicable VAT. These are allowance comparisons; required features, seats, content limits and peak usage can change the tier. External fees and human work are additional inputs.

Basic chatbot for a small or mid-sized business

For the small-shop example, 100 × 3 + 50 = 350 credits. That fits Basic’s 500-credit allowance at a one-credit rate. At two credits per reply, the same customer and testing workload uses 700 credits, exceeding Basic and fitting Essential’s 3,000-credit allowance at $99/month.

This assumes one operator and up to 300 Knowledge Base articles on Basic. Removing Quickchat AI branding requires Professional even if usage stays low. An order lookup also needs a working store API, appropriate access and tested handling of missing or incorrect order details. Include that setup work and any store-provider fees.

AI chatbot for a mid-market company

For an existing helpdesk, 1,000 × 6 + 500 = 6,500 credits. Professional covers that allowance at a one-credit rate, with three team seats. At two credits per reply, usage becomes 13,000 credits and exceeds Professional. REST API access or full conversation exports require Business regardless of this usage calculation; check whether a supported channel integration meets the actual workflow.

Keep the existing helpdesk licence in a total operating-cost comparison. For an incremental budget, count only additional seats or charges caused by the change, and state that boundary. For example, if 300 conversations need eight minutes of human follow-up each, that is 40 staff hours. At an assumed loaded rate of $30/hour, it represents $1,200 of staff time. These are planning assumptions, not a predicted handoff rate or a cash saving. Count AI replies generated before a handoff in the credit budget.

Enterprise generative AI chatbot

For higher-volume support, 5,000 × 4 + 1,000 = 21,000 credits at one credit per reply. That fits Business’s 30,000-credit allowance, subject to its ten seats, 2,000-article limit and other requirements. At two credits per reply the same workload uses 42,000 credits, exceeding the largest published self-service allowance.

For usage above that allowance, discuss a scoped Enterprise agreement. Do not multiply conversations by the advertised resolution rate: the minimum commitment and definition of a billable resolution must be agreed. Website-only Agents cannot assume the extra-credit packs offered to eligible Discord-connected Agents. Model peak periods and human coverage before rollout.

What makes a chatbot more expensive?

Usage is one part of the budget. Required behavior, access to other systems and the work needed to operate the service determine the rest.

The type of bot

Compare the same required behavior before comparing implementation prices.

These categories overlap; adding generative AI does not establish a universal minimum development price. For the distinction between a general-model interface and a business agent, see GPTs vs Quickchat AI: what’s the difference.

Advanced features

Check whether a required capability is included, needs a higher tier or requires separate implementation.

Quickchat AI includes multilingual conversations and AI Actions on all plans. Voice input requires a paid plan. Included functionality still needs configuration and testing.

Integrations

Estimate the work for each connection from its actual scope.

A listed integration does not establish that every workflow is included or needs no engineering. For WhatsApp, Meta’s pricing page describes delivered-message billing by market and category. Check the applicable rate card and free-message conditions separately from Quickchat AI credits and any provider fee.

Branding and user experience

Budget design and conversation work from the screens, flows and accessibility requirements you need. In Quickchat AI, removing vendor branding starts with Professional. A custom widget or interface may require separate development. See the white-label chatbot tutorial for the supported branding controls.

Platform, in-house team, or agency

Compare a platform subscription, an internal implementation and an agency quote over the same period.

• Platform: include setup, knowledge preparation, connected services and evaluation alongside the subscription.
• In-house team: estimate engineering, design, support and review hours using your own loaded rates. Include the work displaced by the project.
• Agency: request deliverables, acceptance criteria, ownership terms, change-request rates and ongoing support costs.

Use scoped hours and rates instead of a generic salary or project-price range.

Support, maintenance, and hosting

Separate infrastructure supplied by the platform from work your team still owns.

A platform may include hosting and model operation while your team maintains its content and workflows. Updating a Knowledge Base is different from training a model; identify the actual task before budgeting for it.

Compliance and security

Ask the responsible legal and security teams to scope the required assessment, contract review, access controls, retention and supplier checks. There is no universal annual GDPR or HIPAA recertification fee to add to every chatbot. Requirements and costs depend on the deployment, data and applicable obligations. Use a scoped estimate from the relevant specialists.

Chatbot pricing models explained

Vendors price chatbots in a few different ways, and each affects your upfront and ongoing cost differently.

Subscription plans

A recurring fee buys a defined bundle of features and allowances. Check commercial-use rights, credits, seats, content limits, exports and the billing period. A low-volume deployment can still require a higher tier for a specific feature. The Quickchat AI plan table below lists the published allowances.

Usage-based pricing

Usage can mean AI replies, tokens, conversations, outcomes or resolutions. Record the exact billable event, included allowance, reset period and charge above the allowance. A vendor’s billing definition of a resolution is not automatically your team’s measure of a successfully solved problem. See AI agent pricing models compared for a comparison of the units.

Hybrid pricing

A contract can combine setup work, a recurring platform fee and usage charges. Request each line separately, including minimum commitments, implementation deliverables and the treatment of testing. Confirm which costs recur when the workflow changes.

Custom builds

Estimate interfaces, knowledge retrieval, integrations, permissions, testing and support operations, then add model/API usage, hosting and maintenance. Agree ownership and exit rights in the contract; paying for a custom build does not by itself determine ownership of every component.

Is it cheaper to build or buy a chatbot?

Compare both options against the same question set, knowledge sources, languages, channels and human handoff requirements. A prototype that generates replies is a different estimate from a service your team can operate for customers. Use the same operating period and expected volume when comparing total costs.

Comparing subscription platforms

These vendor pages were checked on November 5, 2026. Compare the billable units and your applicable contract before comparing headline prices.

Ask each vendor to price the same workload, required features and operating period. Keep reply-metered usage separate from resolution-metered outcomes; one conversation can generate several replies and may remain unresolved.

Building from scratch

Estimate engineering work from your existing systems and team capacity. Include interfaces, retrieval, integrations, permissions, testing and the support workflow, then add recurring model/API usage, storage, hosting and maintenance. Request a quote that separates implementation from ongoing support.

For a worked example of a custom customer-service build, see how to build an AI agent for customer service similar to Klarna.

Build or buy: a quick decision guide

A subscription may reduce implementation work. A custom build may fit requirements the platform does not meet or reuse systems you already operate. Compare measured scope and total costs over a defined period before deciding.

We built an AI chatbot on the free plan: what it really cost

The following walkthrough used the Quickchat AI Free plan for a fictional, non-commercial demo. Free currently includes 50 monthly AI credits and up to 50 Knowledge Base articles. Commercial use requires a paid plan, including a customer-support agent for a real shop.

We created a support agent for a fictional small online plant shop, Fern & Petal. Here is what each step cost.

Step 1: create the agent (free)

Creating the agent, naming it, and writing its instructions is free and takes a couple of minutes with no code. This is the agent’s identity and main prompt.

Step 2: load its knowledge (free)

We added a few help-center articles: shipping and delivery, returns and refunds, and plant-care basics. The Free plan includes up to 50 Knowledge Base articles. Importing content does not spend AI credits, but the article and per-import page limits still apply.

Step 3: add an AI Action (free)

We added one AI Action so the agent can look up an order’s status mid-conversation. AI Actions let the agent call an external API and are available on every plan, including Free for non-commercial use. Running an Action does not itself spend AI credits; external services may charge separately, and the agent’s replies still use credits. For setup steps, see build a free AI agent that takes actions.

Step 4: what actually counts, the AI replies

Each reply on the Standard model spends one AI credit. Other models available on paid plans use one, two or four credits per reply. A conversation can contain several replies. Visitor messages and replies typed by a human operator in the Inbox do not spend AI credits.

There is a catch worth knowing before you pick a plan: testing your own agent spends the same credits a customer would. Every time you try it out, that reply counts. After building Fern & Petal and testing it a handful of times, the account had used 14 of its 50 monthly credits, most of them from our own test messages.

Calculate AI credit use from your expected AI reply counts and selected models, then compare the total with a credit budget.

What Quickchat AI actually costs

Commercial use starts with Starter. The following prices were checked on November 5, 2026, use monthly billing and exclude applicable VAT. The allowance is AI credits, not a fixed number of conversations or replies.

Estimate monthly credits from the replies each model will generate, then add testing:

Monthly credits = sum(customer AI replies per model × credits per reply) + testing credits.

The AI credit documentation lists current rates of 1, 2 or 4 credits per AI reply. For example, 1,000 website conversations averaging six AI replies produce 6,000 replies before testing:

• At 1 credit per reply, customer replies use 6,000 credits. Professional includes 9,000 credits for $299/month.
• At 2 credits per reply, customer replies use 12,000 credits. Business includes 30,000 credits for $999/month.
• At 4 credits per reply, customer replies use 24,000 credits. Business includes 30,000 credits for $999/month.

These examples use declared assumptions rather than observed customer averages. Allow for testing and peaks, and check Knowledge Base limits, team seats and required features. Monthly credits reset each billing cycle without rolling over. If no credits remain, new AI replies pause until credits are available again; human operators can still reply from the Inbox.

Automatic URL refresh, voice input and custom widget translations require a paid plan. Removing Quickchat AI branding starts with Professional. REST API access and full conversation exports require Business. These feature requirements can determine the tier even when reply volume is low.

Eligible Discord-connected Agents can buy packs of 500 extra credits for $5 excluding applicable VAT. These packs are not a general top-up option for a website-only Agent. External channel providers and services called by Actions may have their own fees.

Annual billing is advertised at roughly 17% less than paying monthly. Check the current pricing and checkout total before buying.

Hidden chatbot costs people miss

Include these lines explicitly in the estimate:

• Testing and peaks. Customer replies are only part of the credit budget. Add test runs and a demand scenario based on your own seasonal pattern.
• External providers. Channel, store, CRM and API providers can charge separately, even when the Quickchat AI Action itself consumes no credits.
• Human work. Handoffs, incorrect answers, recontacts and quality review require time. Include them when comparing a chatbot with your current service.
• Implementation and maintenance. Source cleanup, integration changes and access reviews continue after launch. A custom build also needs an infrastructure budget.

For a method of evaluating support costs, see how to reduce customer support costs.

How to estimate your monthly chatbot cost

Estimating your budget is a matter of working through your real needs rather than guessing. If you want a quick number first, the free chatbot ROI calculator estimates monthly savings from your ticket volume and support costs. Then work through these six steps.

1. Define scope and objectives. Decide what problems the bot will solve, who it is for, what tasks it must perform, and how you will measure success. Aim for a clear list of what it must do.

2. Pick a bot type and pricing model. Rule-based, AI-powered, or generative AI, and subscription, usage-based, hybrid, or custom. This gives you a preliminary cost for the core platform.

3. Estimate user and agent counts. How many customers will use it monthly, and how many internal seats will manage it. Seat counts drive subscription tiers on some platforms.

4. Forecast conversation volume. Estimate conversations, AI replies per conversation, model rates and testing usage, including peaks and growth. For Quickchat AI, convert those replies into credits before comparing allowances. For per-resolution pricing, check the vendor’s definition of a billable resolution.

5. List integrations and advanced features. Check which tier includes each required feature, along with provider subscriptions, seats and usage fees. For a custom build, estimate each integration and its ongoing maintenance.

6. Add ongoing work and a peak scenario. Use your observed demand range and implementation uncertainty to choose headroom. Include maintenance, human support, content review and possible tier upgrades.

What if the website has 50,000 visits a month? Measure how many visits start an agent conversation. At an assumed 2% conversation-start rate, 50,000 visits produce 1,000 conversations, which is the helpdesk workload above. The 2% is an invented planning input, not an expected conversion rate. Then estimate replies and testing separately.

For a 12-month total cost of ownership, add one-time implementation to the sum of the year’s platform, provider, human-support and maintenance costs. Use the actual monthly or annual subscription commitment, count it once, and model seasonal demand. Record whether existing staff and helpdesk costs are allocated in full or only incremental costs are included.

For example, 800 conversations with four AI replies each require 3,200 credits at a one-credit rate, before testing. That already exceeds Essential’s 3,000-credit allowance. More capable models can consume additional credits, and taxes or external provider charges can add to the subscription. Calculate those inputs before choosing a tier.

Do chatbots pay for themselves?

Payback depends on measured costs and outcomes over the same period. A reduction in workload can free staff time without reducing payroll. Record that capacity benefit separately from a cash saving.

Cost per chat: bots versus humans

Compare comparable cases, including escalations and repeat contacts. Dividing a subscription by all conversations produces a platform cost per conversation, not a total cost per resolved case. For a total operating-cost measure, include allocated platform, provider and human costs, and state how you verify resolution and handle recontacts.

Revenue, not just savings

Measure qualified leads or incremental contribution margin with an appropriate comparison. A chat-assisted sale is not proof that the chatbot caused the sale. Avoid counting the same benefit as both a revenue gain and a cost saving.

A simple ROI formula

Use the same period and cost boundary for benefits and costs:

ROI (%) = (Benefit before chatbot costs - Total chatbot costs) / Total chatbot costs x 100

For a hypothetical annual example, $18,000 of verified benefit before chatbot costs and $12,000 of total costs gives 50% ROI: ($18,000 - $12,000) / $12,000 × 100. These are invented inputs to demonstrate arithmetic, not expected results. Keep the result provisional until the benefit and cost inputs are supported. See how to calculate chatbot ROI for the measurement process.

How to lower your chatbot costs

• Start with a defined use case. Test representative questions and escalation before expanding the workflow.
• Evaluate a platform with your requirements. A personal, non-commercial trial can test behavior; a real business deployment requires commercial-use rights.
• Reuse existing APIs. Verify permissions, limits and maintenance responsibilities before commissioning a new connection.
• Match model and plan to evidence. Compare answer quality and credit use on the same test set, then check the required features and peak allowance.
• Scope larger deployments. Request an applicable usage agreement instead of assuming a public entry rate applies to all traffic.
• Reduce unnecessary exchanges. Improve source content and clarify instructions while preserving answer quality and access to a human. See how to scale customer support efficiently.

Frequently asked questions

How much does an AI chatbot cost per month?

Quickchat AI’s Free plan is for personal, non-commercial use and includes 50 monthly AI credits. Commercial plans start with Starter at $9/month, with self-service plans up to Business at $999/month, excluding applicable VAT. Estimate AI replies, the selected model’s credits per reply, testing and required features before choosing a tier. Enterprise pricing starts at $0.50 per resolution with a minimum of 2,000 resolutions per month.

Is there a free AI chatbot plan?

Quickchat AI’s Free plan is for personal, non-commercial use and includes 50 monthly AI credits, up to 50 Knowledge Base articles and one user. Commercial deployments require a paid plan, starting with Starter at $9/month excluding applicable VAT. Check feature and usage limits before selecting a plan.

How much does it cost to build your own chatbot from scratch?

Request an estimate for the requirements you actually need: interfaces, knowledge retrieval, integrations, permissions, testing and support operations. Add recurring model/API usage, storage, hosting and maintenance. A prototype and a customer-facing support service can have very different costs, so a general development-price range is not a quote for your project.

What makes a chatbot more expensive?

The biggest drivers are reply and conversation volume, the AI model you use, the number of channels, the integrations and AI Actions you connect, and time spent on human handoff. If you build your own, engineering and ongoing maintenance dominate the cost. On Quickchat AI self-service plans, usage is measured in AI replies multiplied by the selected model’s credit rate; required features also affect the plan.

How much does a chatbot cost for 1,000 conversations a month?

First estimate AI replies per conversation and the model’s credits per reply. At six AI replies per conversation, 1,000 conversations use 6,000 credits at a one-credit rate, 12,000 at two credits, or 24,000 at four credits, before testing. Compare the total with current allowances and required features. Conversation count alone cannot determine a plan or an all-in price.

Does testing the chatbot cost extra?

On Quickchat AI self-service plans, AI replies generated during testing spend credits at the selected model’s rate, just like customer-facing replies. Include testing in your monthly budget. Enterprise advertises unlimited testing; confirm the applicable agreement. Other vendors use different billable units and testing rules.

Are AI chatbots more expensive than rule-based bots?

AI adds model usage and evaluation work, but the total cost depends on scope and what you can reuse. A complex rule-based workflow may require more implementation work than a narrow AI use case. Compare the same required behavior, operating period and support workload rather than assuming a universal minimum development price.

Is it cheaper to build or buy a chatbot?

Compare both options against the same required channels, knowledge sources, languages, human handoff and evaluation results. For a subscription, include the plan and external provider costs. For a custom build, include engineering, model/API usage, hosting and ongoing maintenance. Which costs less depends on your scope, existing systems and operating period.

How much does a chatbot cost per conversation?

A conversation is not a standard billing unit across platforms. For Quickchat AI self-service plans, estimate the AI replies in each conversation and multiply by the selected model’s credit rate, then include testing. Enterprise starts at $0.50 per resolution with a minimum of 2,000 resolutions per month, subject to the agreed scope. Divide total operating costs by verified outcomes only when those costs and outcomes cover the same period.

What hidden fees should I watch for in chatbot pricing?

Watch for overage or per-resolution charges above your plan allowance, the fact that testing consumes your allowance, charges for premium support or specific integrations, setup and onboarding fees, extra seats, and data or analytics retention limits. If you build your own, the easy-to-miss costs are ongoing maintenance, model API bills, and retraining.

Conclusion

For Quickchat AI, start with the AI reply volume and model credit rate, then check commercial-use rights, Knowledge Base limits, seats and required features. Include testing, applicable tax and external provider costs. Compare a custom implementation against those same requirements and a defined operating period.

Measure usage on representative questions before committing to a rollout. The business case for chatbot adoption covers how to evaluate costs and benefits. You can create an account for a personal, non-commercial trial or talk to us about a commercial deployment.