Pricing units

Message credits, conversations, and AI resolutions are not the same thing.

The price that looks cheapest at signup can become confusing when one vendor meters replies, another meters support conversations, and another meters solved AI cases.

Editorial illustration of different AI chatbot pricing meters, including message credits, conversations, resolutions, seats, and add-ons.

Short answer

Do not compare AI chatbot plans by monthly price until you know the unit that runs out first. A message credit is usually not a visitor, a conversation is usually not an AI reply, and an AI resolution is closer to a solved support case than a raw chat message.

FastBots and Chatbase are easier to model when the main job is trained website answers, because the first question is message credits and source capacity. Tidio and ChatBot.com can make more sense for support operations, but the buyer must model human replies, AI conversations or resolutions, seats, and workflow scope.

This guide explains the units. For the next layer, use the pricing-traps checklist, the scaling-cost guide, or the small-business plan picker.

Comparison method

Translate every plan into the same buyer question.

The practical question is not which plan has the lowest headline price. It is which unit grows first for your site.
01 Unit

Name the thing being metered

Credits, conversations, resolutions, seats, and actions are different units. Translate the plan into the job your site needs.

02 Burn

Estimate how fast it runs out

A visitor might ask one question, trigger several AI replies, hand off to a person, or require a resolved support case.

03 Gate

Find the first upgrade pressure

The first jump may be model cost, source size, live chat, AI actions, users, branding removal, or refill behavior.

04 Fit

Choose by workload, not sticker price

A lightweight FAQ bot and a customer-support workspace can both be good buys, but their cost math is not the same.

Unit map

Four pricing units buyers often mix together.

These are not perfect universal definitions. They are the practical buyer meanings from the official pricing surfaces checked for this article.

Unit Where it appears What it means Buyer check
Message credits FastBots, Chatbase Usually a budget for AI model replies or model requests inside the chatbot product. One visitor conversation can consume several replies, and some models or action workflows can spend more than one credit per answer. Estimate monthly bot replies and preferred model choice before treating the credit allowance as traffic capacity.
Billable conversations Tidio Customer Service A support conversation can be billed when a human agent replies or starts the conversation. Human-handled support volume can grow separately from AI automation volume, so a busy inbox can hit the support meter first. Separate unanswered visitor messages, human replies, Lyro AI replies, and automation flows before choosing a quota.
AI or Lyro conversations Tidio Lyro AI Agent A customer interaction that receives at least one reply from the AI agent. AI support can become its own quota even when live-chat conversations or Flows automation are priced separately. Model AI-handled questions separately from live-chat seats, tickets, and rule-based Flows reach.
AI resolutions ChatBot.com by Text A support outcome where the AI agent solves at least one customer question in a conversation. Resolution allowances, users, AI agents, API calls, workflows, and automatic refill packs can all affect the real monthly bill. Ask what counts as a resolution, what is excluded, and what happens when the included monthly allowance is reached.

Tool fit

Which pricing meter should you model first?

Use this as a unit-mapping aid, not as a final ranking or hands-on billing result.

Model replies

FastBots

Message credits

Start here if

Small sites that want trained answers from website content, files, links, and simple lead-capture workflows.

Before you choose

The official pricing page showed 50 free credits, then 2,000, 5,000, and 10,000 monthly credits on the self-serve paid plans. Standard models can use 1 credit per reply; advanced models can use more.
Check FastBots

Sources and actions

Chatbase

Message credits plus add-ons

Start here if

Source-heavy sites that need trained answers, AI Actions, agent limits, source limits, and workspace controls.

Before you choose

The pricing page showed 50 free credits, then 500, 4,000, and 15,000 monthly credits on Hobby, Standard, and Pro. Actions can make multiple model requests, and auto-recharge is a separate credit behavior to check.
Check Chatbase pricing

Support workload

Tidio

Conversations, Lyro, and Flows

Start here if

Teams that want live chat, tickets, automation, Lyro AI, and human support in one customer-service workspace.

Before you choose

The pricing page separates billable conversations from Lyro AI conversations and Flows visitors reached, so price human support, AI support, and automation reach separately.
Check Tidio pricing

Resolved support cases

ChatBot.com

Users plus AI resolutions

Start here if

Support teams that want an AI agent inside the broader Text workspace with live chat, help desk, workflows, and API scope.

Before you choose

The pricing page showed per-user plans with included AI resolutions, plus automatic 50-resolution refill packages when the limit is reached.
Check ChatBot.com

Workload first

Choose the meter that matches the job.

A chatbot that answers public website questions and a chatbot that resolves support cases can both be useful. The mistake is forcing both into the same cost model.

Message credits and source size

Website FAQ bot

The buyer usually needs enough replies, crawl pages, files, and training data before support-workflow features matter.

Check: Start with FastBots or Chatbase-style credit math, then check handoff only if the bot must route leads.

Billable conversations and seats

Live chat with human replies

The cost can grow when a person replies, when more staff need access, or when inbox/ticket features become required.

Check: Use Tidio-style conversation math and separate human replies from AI conversations and Flows reach.

AI conversations or AI resolutions

AI support deflection

The buyer is paying for solved or AI-handled support workload, not just for a fixed number of chat messages.

Check: Ask how the vendor counts solved issues, what is excluded, and what happens after the included quota.

Form fields, handoff, actions, and CRM routing

Sales or quote intake

The chatbot may answer a few questions but still need safe handoff, lead details, scheduling links, or action gates.

Check: Use the quote-request and appointment-booking guides before assuming a low-price plan covers the full workflow.

Source basis

Official pages checked on May 27, 2026

This guide uses current public pricing pages to identify usage units, included allowances, and overage/refill behavior. It does not claim that ChatbotEdge bought every plan, tested invoice behavior, or simulated live support volume.

Are message credits the same as conversations?

No. Message credits usually meter AI replies or model requests. A visitor conversation can use several replies, especially when the chatbot asks clarifying questions or uses a higher-cost model.

Are AI resolutions the same as AI replies?

No. A resolution is a support outcome. On the checked ChatBot.com pricing page, a resolution is tied to the AI agent solving at least one customer question in a conversation.

What should I compare first?

Start with the job. For website answers, compare message credits and source limits. For support, compare billable conversations, AI conversations or resolutions, users, seats, and handoff.

Next checks

Compare the unit before the plan.

Once the pricing unit is clear, check scaling costs, plan gates, source limits, handoff, and quote or booking workflows before you commit.