Botpress Pricing Plans: Free, Plus and Team for a Small Business

Botpress Free costs $0 and includes 100 conversations; Plus is $150 a month billed annually with 250 conversations a month; Team is $750 a month billed annually with 1,500. Plus adds 100-conversation packs for $65, and Team packs cost $50. AI model usage is another meter. For a small firm, choose by measured conversations, handoff needs and total monthly cost.

The first trap is treating a conversation as a website visit. Botpress counts an exchange with at least two end-user messages in the billing month. The second is assuming a 250-conversation Plus plan simply stops at 250: its pricing page describes automatic packs on paid plans near the quota. The third is ignoring model usage, which has its own included allowance and recharge settings.

Follow me on Instagram@sagnikteaches

Three plan prices, three different jobs

Botpress's current pricing page presents Free, Plus, Team and a custom Enterprise option. Free is an experiment or low-volume starting point: 100 conversations, three seats and three AI agents, with no top-ups or overages. Plus is aimed at a small operating team: $150 monthly on annual billing, 250 conversations a month, three seats, unlimited AI agents, white-label webchat and $25 of included AI usage. Team is $750 monthly on annual billing with 1,500 conversations, unlimited seats, routing and assignment, team-level analytics and $150 of AI usage.

Connect on LinkedInSagnik Bhattacharya

The comparison table lists Free's 100 included conversations without the “/mo” label that Plus and Team carry, and it shows no additional conversations and no auto-recharge for Free, while the page's FAQ defines a conversation by billing month. For a Free account, confirm its actual reset date in the workspace before assuming exactly when the next 100 become available. Plus and Team figures are explicitly monthly on the live card.

Subscribe on YouTube@sagnikteaches

The prices above are annual-billing monthly equivalents; the page says annual billing saves 20%. Do not treat $150 as the price of one cancellable month without checking the monthly toggle and checkout. Likewise, the $750 Team price is not a recommendation that a shop with 1,500 conversations must buy Team. A small group might remain on Plus with packs if it does not need Team's routing and roles. Compare the complete invoice, not just the included-volume headline.

What Botpress calls a conversation

Botpress's pricing FAQ says an exchange counts after at least two messages from the end user within the billing month. An anonymous visitor who opens three separate sessions may generate three conversations. A thread that spans two billing months can count in both. AI-only and human-assisted exchanges count the same. For voice, the pricing page says three minutes count as one conversation, although its comparison table lists the voice channel only on Enterprise. These details matter if a website bot invites visitors to ask several follow-up questions or if staff take over a chat.

For an illustrative interior designer, 1,000 people visit a portfolio in a month, 180 open the chat and 70 ask at least two messages. The conversation count is closer to 70 than to 1,000 or 180, subject to repeated anonymous sessions and the precise dashboard count. Free's 100 could support the pilot. If an automatic greeting and one customer reply do not meet the two-end-user-message definition, do not count them as full conversations in your forecast; verify the workspace meter instead of trying to infer it from raw visitor analytics.

A second illustrative case is a security installer whose customers come back to the website bot several times to check an appointment. If 80 people each start two anonymous sessions that qualify as conversations, volume may be around 160. The owner might think “only 80 customers” and choose Free, then hit the allowance. Give the bot a clear human contact route and test how repeat visits appear in Botpress reporting before sizing a plan.

For an illustrative solar installer, long conversations may be useful: the bot collects roof type, approximate bill and preferred contact method. The firm should still avoid requesting documents or sensitive account details before deciding how that information is stored and handed over. A long, qualified conversation can count as one conversation while using more AI model resources than a short FAQ exchange. That is why the conversation meter alone does not settle the price.

A 400-conversation month on Plus

Assume an illustrative five-person signage company runs a website bot for order status, artwork requirements and quote enquiries. Its pilot projects 400 qualifying conversations in a month, with three staff maintaining the bot. Plus includes 250. The shortfall is 150, so two 100-conversation packs are needed if use reaches 400. At the listed $65 each, that is $130 in packs; combined with the $150 Plus base, the planning total is $280 for that month before any extra AI usage or other services.

Botpress says paid plans automatically add a 100-conversation pack when usage reaches 95% of the allowance. The first pack may therefore appear before the 250th conversation, and later packs may be triggered as available allowance is approached. Its page says those packs persist month to month until removed in billing settings, but unused conversations expire at month end. Check the live billing view before and after a busy period. A quiet following month could still retain an unneeded pack setting.

At 400 conversations, Plus plus packs is far below Team's $750 list price. Team may still be warranted if the firm needs formal routing to several support groups, role-based access, unlimited seats or team analytics. If three people can manage the bot and one shared handoff queue suffices, volume alone does not justify Team in this example. The Botpress versus Chatbase comparison can help if the team questions whether Botpress's builder is more than it needs.

Because each pack is added at 95% of the current allowance, the pack count runs slightly ahead of simple division at higher volumes. At 1,000 conversations, Plus has added nine packs: the ninth arrives at about the 998th conversation, 95% of a 1,050 allowance. Packs then cost $585 and the month $735, just under Team's $750. From about the 1,093rd conversation a tenth pack lands and Plus reaches $800. These are pack-trigger comparisons, not a guarantee of the exact invoice, because retained packs and model usage affect it. Team can become attractive before that arithmetic crossover if its workflow controls are needed.

Check the bill at quiet, normal and busy volume

Use three cases instead of a single monthly average. The illustrative signage company has 180 qualifying conversations in a quiet month, 400 normally and 620 during its busiest campaign. On Plus, the quiet month needs no extra 100-conversation pack on the simple allowance calculation, so the listed base remains $150. The normal month needs two packs and reaches $280. The busy month exceeds 250 by 370, so simple division says four packs, but the 95% trigger adds a fifth at about the 618th conversation (95% of a 650 allowance). Five packs add $325 and the total becomes $475. Model usage can change each figure; the table of conversations alone does not predict it.

That range is more useful than announcing “Botpress costs $150”. If the company can absorb the busy month, it may have no reason to pay $750 for Team. If it needs Team's routing every month, the higher fixed cost buys operating controls even in the quiet month. If a short promotion causes the spike, ask whether the packs can be reviewed promptly afterwards. Botpress says packs persist in billing settings until removed even though unused conversations expire at month end, so somebody must own that check.

Now change the workload instead of the traffic. Suppose half the 400 monthly conversations are simple FAQ answers and half involve a long design brief, repeated clarifying questions and several model calls. An average AI usage figure from ten simple FAQ tests will understate the bill. Sample both categories and weight them by forecast volume. If simple chats average an illustrative $0.04 and longer chats average $0.16, 200 of each imply $8 plus $32, or $40 model usage. Plus includes $25, leaving an illustrative $15 above the included amount. Botpress covers that by buying $10 credit grants automatically once AI spend reaches 95% of the limit, so this month would see two grants, $20 charged, with about $5 carried forward for up to 12 months. The arithmetic is transparent, but the unit rates must come from your own pilot.

Include abandoned and unhelpful chats in the review. A visitor may ask two questions, receive a wrong answer and leave. That may still count as a conversation without producing a useful enquiry. If 120 of the 400 conversations end that way, buying a larger quota does not solve the problem. Inspect a sample of those transcripts, correct missing knowledge and retest. Price per useful resolved enquiry is a better owner metric than price per raw conversation, provided you define “resolved” consistently and do not let the bot mark its own incorrect answers as successes.

The separate AI usage balance

Botpress says Plus includes $25 of AI usage and Team $150. It explains this as $0.10 of model usage per included conversation. That is an allowance against the cost of calling a language model, not a promise that every conversation costs exactly ten cents. A bot that retrieves a short answer and makes one model call may use less; a bot that repeatedly searches a large knowledge base or rewrites responses may use more. The workspace's actual AI usage meter is the evidence.

In the illustrative signage company's 400-conversation month, suppose the measured average model spend during the pilot is $0.08 per conversation. Multiplying gives $32 of model usage. Plus includes $25, so an illustrative $7 remains beyond the included amount, which one automatic $10 credit grant would cover. If the average is $0.18, 400 conversations imply $72 model usage and $47 beyond the included $25; with each grant bought at 95% of the running limit, that month would trigger six $10 grants, $60 charged, with the unused $13 carried forward. The base subscription and conversation packs remain $280 in both cases; AI usage changes the all-in figure.

Do not infer model spend from total conversations by using the included $0.10 as a fixed rate. Measure at least 20 varied interactions: a routine opening-hours question, a complex order-status query, a failed retrieval, a human handoff and a visitor who asks several follow-ups. Note both conversation count and AI usage before and after. If one scenario is much more expensive, shorten its knowledge context, reduce repeated model calls or route it to a person. A human-assisted conversation still counts as a conversation, but its model usage can differ.

For an illustrative interior designer, a bot might answer “What is the minimum project size?” from a short approved FAQ. The cheaper path is one accurate answer with a link to contact the team. Asking the model to generate a personalised design concept for every visitor uses more model resources and could make promises the firm has not reviewed. Keep the first bot narrow. The beginner FAQ bot tutorial is a suitable build path when the goal is to answer approved questions.

Free, Plus or Team: decide with operational thresholds

Free fits a controlled pilot when forecast qualifying conversations are comfortably below 100 and the owner can tolerate a hard limit. Botpress's comparison table lists no additional conversations and no auto-recharge on Free, so a public bot close to that limit needs a fallback. The fallback could be a clearly visible contact form or a human email route; it should not silently fail. Test from a new browser session after the allowance is nearly spent and confirm what the visitor sees.

Plus fits a small team that wants more than a test, needs its white-label webchat, WhatsApp channel or live-chat technical support and can track the automatic packs. A five-person office with only two or three bot administrators can still fit Plus's three seats, provided the other colleagues do not need workspace access. If every customer support colleague must edit, route or report on conversations, Team's unlimited seats and routing features may matter. Do not share one login simply to remain under a seat count; named access makes changes and handoffs easier to audit.

Team fits an organised support operation with several queues, routing rules and team-level reporting. For an illustrative security installer, the bot might sort new installations, service calls and existing-contract issues to separate people. If misrouting has real cost, Team's routing and assignment could be more important than its 1,500 included conversations. The owner should demonstrate a handoff from the bot to the right group and check who sees the transcript. If the same three-person office handles every issue, the extra control may add complexity without benefit.

Use a simple monthly decision sheet: qualifying conversations; peak-day and repeat-session pattern; AI usage amount; packs added; staff who need access; handoffs that failed; customer enquiries resolved or converted. A plan upgrade is justified when it solves a measured constraint. An illustrative bot that has 90 conversations but gives wrong order-status answers needs better source data and testing, not Team. A bot that gives accurate answers but regularly reaches its Free cap may need Plus or a more limited public scope.

Five test conversations before a paid rollout

First, ask a basic approved question, such as “What artwork files do you accept?” The expected answer should match the firm's written policy and link to the right human route if the file is unusual. Second, ask a two-part question about a deadline and custom size. The bot should request missing details rather than inventing a delivery promise. Third, ask for an order update without providing an order number. It should not reveal another customer's information or imply that it checked an order it cannot access.

Fourth, ask to speak to a person. Record whether the handoff is visible and whether the human receives enough context to avoid making the customer repeat everything. Fifth, begin a conversation near the plan limit and check the visitor's experience if the allowance or AI usage limit is reached. A bot that goes silent can lose the enquiry even though its dashboard shows a neat usage chart. The silent-bot checks can help distinguish plan limits from configuration problems.

Keep an issue log with prompt, expected response, actual response, severity and owner. In an illustrative signage test, the bot says “We accept any image file” while the actual policy accepts only specified formats above a resolution threshold. Correct the knowledge source and retest the same prompt. In another test, it directs a customer to a booking page that no longer exists. Fix the link and inspect the published bot, not only the builder preview. These defects affect the buying decision because a cheap bot that misinforms customers is expensive to operate.

Review the bill after the pilot. Compare measured conversation count with the forecast, inspect model usage, check whether an automatic pack remains enabled, and decide who maintains the knowledge base. If the bot has 60 useful conversations and no capacity problem, stay on Free while it remains available. If 400 conversations are mostly resolved accurately, the Plus calculation gives a credible budget. If routing and several human teams are essential, evaluate Team against those operational requirements, not against the word “business” in a plan name.

Botpress billing questions for a small team

Does one visitor message count as a Botpress conversation?

Botpress defines a billable conversation as an exchange with at least two messages from the end user in the billing month. Multiple anonymous sessions can each count separately, and a conversation spanning two months can count in both. Use the dashboard's measured count rather than simply counting website visitors or every inbound message.

Can I turn off automatic conversation packs on Plus?

No. Botpress's pricing FAQ says auto-recharge for conversation packs cannot be switched off, because pausing conversations would pause the product. A pack of 100 is added at 95% of the quota, with warnings at 75% and 100%. Packs persist month to month until removed in billing settings, while unused conversations in a pack expire at month end, so review billing settings after every busy month.

Are AI model charges included in the monthly price?

Plus includes $25 of AI usage and Team includes $150, described by Botpress as $0.10 per included conversation. Model calls are a separate meter. On paid plans, when AI spend reaches 95% of its limit, Botpress automatically buys a $10 credit grant at cost, and unused grant balance carries over for up to 12 months. Track both meters in a pilot.

Further reads

• Roofing Chatbot Platforms With AI Compared: Qualifying Storm Enquiries — No roofing chatbot wins on its vendor page alone: the ten questions a storm bot must ask, nine platforms compared, and a ten-enquiry test to run first.
• Bolt.new Pricing: How Many Tokens a Business Tool Really Uses — Price a small business tool in Bolt by measuring prompts, fixes and publishing, rather than guessing a token total.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: Botpress pricing page, plan comparison table and pricing FAQ (checked 5 November 2026).