Key Takeaways
- At 1,000 monthly AI conversations, chatbot costs can run from about $49 a month on the low end to $990 or more, depending on how the vendor bills for usage
- Pricing splits into five billing units – message, credit, query, conversation, and outcome – so the same list price can mean very different real-world costs
- At 5,000 conversations a month, published costs range from around $188 to nearly $5,000, showing how fast outcome-based pricing can climb
- Hidden extras like overages, seats, branding removal, and human handoff fees often add more to the bill than the base plan itself
- Choosing the right pricing model for expected conversation volume often matters more than which vendor name sits on the invoice
Shopping for an AI chatbot usually starts with one question: what does it actually cost? The answer depends less on which brand sits on the pricing page and more on how that brand chooses to bill. A plan that looks cheap at low volume can turn into one of the priciest options once conversation counts climb, while a plan that looks steep upfront can end up being the better deal at scale.
What 1,000 Conversations Really Cost
At roughly 1,000 AI conversations a month – a volume typical of a small service business just getting its chatbot off the ground – published 2026 pricing shows real costs starting around $49 a month and stretching past $990, based on vendor pricing pages. Botsonic’s Professional plan lands at $49 a month for 3,000 messages, which covers 1,000 conversations with room to spare. SiteGPT comes in close behind at $59 a month on its Starter plan. On the far end, Fin, the AI agent built by the company formerly known as Intercom, can run up to $990 a month before any seat costs, because it charges per resolved outcome rather than per message or plan tier.
That spread between $49 and $990 for the exact same volume of conversations is the whole story in miniature. Two chatbots can serve the same number of customers, answer the same kinds of questions, and still land on bills that differ by nearly 20 times, purely because one vendor counts messages and the other counts resolved outcomes. Business owners evaluating options should treat the sticker price as a starting point rather than the final answer, since the real number depends heavily on how usage gets measured. MediaDrive AI works with service businesses managing this kind of pricing maze, helping them understand what a tool will actually cost once real customer volume hits it.
Small businesses testing the waters have other entry points worth knowing about too. Botsonic’s Starter plan, at $19 a month billed monthly or $16 a month billed annually, is the lowest published paid tier in this category and includes 1,000 messages with one chatbot. That makes it a reasonable way to see how an AI chatbot performs before committing budget to a higher tier built for real conversation volume.
Why Pricing Swings So Widely
The gap between a $49 bill and a $990 bill for the same conversation count is not random. It comes down to what unit of work a vendor decides to charge for, and that choice shapes everything else about how costs behave as a business grows.
Five Billing Units: Message, Credit, Query, Conversation, Outcome
Every AI chatbot vendor picks one of five things to count before sending an invoice, and knowing which one applies is the fastest way to compare two quotes fairly.
- Message: Botsonic and SiteGPT both bill per message, where a message is one customer question paired with one AI reply. A single conversation typically runs through several of these exchanges.
- Message credit: Chatbase uses credits, a slightly looser unit that still ties back to how many AI replies get sent each month.
- Query: CustomGPT counts queries, which fits a platform built around searching large document libraries rather than open-ended chat.
- Conversation: Botpress and Tars bill per conversation, counted once a customer sends at least two messages, which bundles several exchanges into one billable unit.
- Outcome: Fin charges only when a conversation ends in a resolution, meaning an unresolved chat costs nothing but a resolved one costs $0.99.
None of these units are equal in size, so a $600 price tag for 1,000 Botpress conversations is not directly comparable to a $19 price tag for 1,000 SiteGPT messages. A single conversation might contain three or more messages, which means converting every quote to a shared, real-world volume is the only honest way to line vendors up side by side.
What Drives the Cost Up or Down
Beyond the billing unit itself, a handful of practical factors push the final number higher or lower for any given business.
- Conversation complexity: Simple FAQ-style exchanges cost less to process than multi-turn conversations that require pulling from large knowledge bases or handing off to a human.
- Feature depth: Branding removal, analytics, multiple chatbots, and extra team seats usually sit behind higher-tier plans.
- Integration needs: Connecting a chatbot to a CRM, helpdesk, or booking system often requires a plan tier built for that level of complexity.
- Billing frequency: Annual billing commonly shaves a meaningful percentage off the monthly rate compared with paying month to month.
- Vendor positioning: A platform built for developers building custom agents, like Botpress, prices differently than a self-serve tool built for quick setup, like Botsonic.
Cost at 1,000, 5,000, and 10,000 Conversations
Looking at a single volume only tells part of the story, since the vendor that wins at 1,000 conversations is not always the one that wins once volume climbs.
At 1,000 conversations a month, Botsonic’s $49 plan and SiteGPT’s $59 plan sit closest together, while Chatbase lands around $150 and Landbot around $160. Fin’s outcome-based pricing can reach $990 a month at this volume if every single conversation resolves.
At 5,000 conversations a month, the order shifts. SiteGPT drops to the lowest published cost at $188 a month, Botsonic rises to $349, and Chatbase climbs to $500. Fin’s ceiling at this volume reaches $4,950 a month before any Intercom seat costs are added, which is roughly 26 times SiteGPT’s figure for the identical volume.
At 10,000 conversations a month, SiteGPT remains the lowest published option at $365 a month, with Botsonic trailing at $524 and Chatbase considerably higher. Fin’s figures assume every conversation ends in a resolved outcome; at a 50% outcome rate, the real bill would land closer to half those ceiling numbers. These figures also come with fine print worth remembering: SiteGPT’s numbers are based on GPT-4.1-mini messages, and the same plans cover one-tenth as many messages if a business opts for the more advanced GPT-4.1 model.
Flat Fee vs. Per-Resolution: Which Model Wins at Scale
The flat-fee model, used by Botsonic and SiteGPT, charges a predictable monthly rate tied to a message allowance. Costs rise in steps as a business moves to a higher tier, but the bill stays flat within that tier no matter how many conversations actually happen. This predictability makes budgeting simple and tends to reward businesses with steady, high-volume traffic.
The per-resolution model, used by Fin, charges nothing for conversations that go unresolved and $0.99 for every one that does resolve. Per-resolution pricing can suit businesses with low or unpredictable volume, since there is no minimum spend tied to a plan tier. The tradeoff shows up at scale: as conversation volume grows and resolution rates climb, the bill grows right alongside it, with no ceiling unless a business negotiates a custom arrangement.
The practical takeaway is straightforward. A business expecting steady growth past a few thousand conversations a month generally comes out ahead with a flat-fee plan, while a business testing a chatbot for the first time, with low and uncertain volume, may find per-resolution pricing cheaper in the early months.
Hidden Charges Beyond the Plan Price
The number on a pricing page rarely tells the whole story. A plan advertised as an entry-level rate can grow substantially once a business adds the features it actually needs to run day to day.
Overages, Seats, and Human Handoff
Running past a plan’s included message or conversation allowance triggers overage charges that vary widely by vendor. Chatbase charges $40 per 1,000 extra message credits, Botpress charges $65 per 100 extra conversations on its Plus tier, and Landbot charges $0.10 per extra AI chat once a plan’s allowance runs out. Seats add another layer of cost for any business that wants a human team member handling escalations: Intercom’s seat pricing for Fin runs from $29 up to $132 per seat a month billed annually, while Zendesk charges $19 to $115 per agent a month on annual billing. Human handoff itself can carry its own fee, such as Botsonic’s $199-a-month Zendesk handoff add-on or Tars’s live chat handoff at $600 a year per seat.
Branding Removal, Setup, and Support Fees
Removing a vendor’s logo or watermark from the chat widget is rarely included in entry-level plans. SiteGPT charges $39 a month billed annually for branding removal, Botsonic charges $49 a month, and Chatbase charges $99 a month, while CustomGPT only includes it starting on its $499-a-month Premium tier. Setup and support fees can add up even faster: Tars lists professional support services at $5,000 a year per agent for businesses that want a fully managed setup. None of these costs show up in the headline price, which is exactly why comparing two chatbot quotes purely by their advertised monthly rate can be misleading.
Matching a Pricing Model to Your Volume
Choosing the right pricing model comes down to answering three practical questions: how many conversations are expected each month, how complex those conversations tend to be, and how much budget predictability matters.
A business handling under 1,000 conversations a month, with a tight budget and simple FAQ-style questions, often does best with a low-cost flat-fee entry plan or a free tier used for testing. A business expecting steady growth into the 5,000-to-10,000 range benefits from locking in a flat-fee plan early, since that structure rewards higher volume with a lower effective cost per conversation. A business with low or seasonal volume, where conversation counts swing month to month, may find a per-resolution model like Fin’s more forgiving, since there is no bill at all for months with little traffic.
Businesses layering a chatbot into a broader customer response strategy should also weigh how the tool fits alongside other follow-up systems already in place, such as missed-call text-back or appointment booking. A chatbot that answers a question well can still leave a lead hanging afterward, leaving part of the problem unsolved.
Pricing Model Matters More Than Vendor Choice
The research bears out a point worth repeating: the billing model a business chooses tends to shape total cost more than which specific vendor gets picked. Two platforms using the same billing unit, like Botsonic and SiteGPT’s shared message-based pricing, end up roughly comparable at low volume and diverge mainly based on tier size and overage rates. A platform using a fundamentally different billing unit, like Fin’s per-outcome pricing, can look cheap at tiny volume and expensive at scale, or the reverse, depending entirely on how many conversations actually resolve.
The practical move for any business owner evaluating options is to estimate realistic monthly conversation volume first, then run that number through each vendor’s actual billing unit rather than comparing sticker prices in isolation. A $59-a-month plan and a $0.99-per-outcome plan cannot be judged side by side without doing that math, and skipping it is how businesses end up with monthly bills far higher than expected.
Getting the chatbot bill right only pays off if those conversations turn into booked business once they end. For service businesses looking to make sure every inquiry, whether it starts in a chat window or a phone call, gets followed up on and converted into an appointment, a lead follow-up system built specifically for that handoff is worth a look. Click the link and reach out to MediaDrive AI for an honest and transparent option.
MediaDrive AI
bill@mediadrive.ai
+1-719-427-5511
8660 Dry Needle Place
Colorado Springs
CO
80908
United States
