Voiceflow pricing plans, credits and chat costs

Voiceflow doesn’t print what its plans cost. Its pricing page offers agencies a free trial and businesses a demo, and its billing docs say the current prices of its four plans, Free, Pro, Business and Enterprise, are shown inside Voiceflow under Plans and Billing. What it does publish is a rate card. On self-serve plans, each message a visitor sends costs $0.005 in credits, the AI model’s tokens are charged on top, and a phone call costs $0.05 a minute. The $60 a month for Pro that many sites quote comes from a Voiceflow post of April 2025. Every figure here was read on Voiceflow’s own pages on 23 September 2026, and each one links there.

Many of the pages that rank for Voiceflow pricing come from voice-agent vendors, and they price agents that answer the phone. This one is about chat, an AI agent that answers visitors on your website, and what Voiceflow costs for that job, with a month of chat priced from its public rates.

Deacon is ours, and it’s built for that job. It’s an AI customer support agent for founders who answer support themselves. It answers your customers from your own docs in seconds, in a chat widget on your site, and shows you what they’re stuck on and what they want next. There are no flows to build, every plan answers with the same model, and nothing is billed past your plan. The prices are on its pricing page, and the free plan needs no card.

Voiceflow’s four plans

Voiceflow’s docs describe each plan by who it’s for and what it adds. They give no price or credit figure for any of them.

The docs say a bill has three parts. There’s a monthly plan fee, add-ons such as editor seats and phone numbers, and usage paid in credits. It’s all billed to your organisation, and one pool of credits is shared by all its workspaces.

Every paid plan includes some editor seats, and more are an add-on, though the docs don’t say how many come with each plan. Viewers are free on every plan. Each workspace gets one Voiceflow phone number, and another add-on raises how many calls can run at once, a limit that doesn’t apply to chat. If you cancel, you keep the paid features to the end of the period and then drop to the free tier.

Free is for trying Voiceflow out. Its credits are “a one-time credit grant that does not renew, so it isn’t suitable for production usage,” the billing docs say. The agencies route on the pricing page offers a “Free trial, no credit card required”. To run a chat agent on your live site month after month, you need a paid plan.

Where the $60 figure comes from

The plan prices on other sites come from Voiceflow’s own post of April 2025, which moved its billing to credits. Its table read like this, with Business then called Teams.

Billed yearly, the smallest Pro bundle was $648 a year. The same post cut the yearly discount from 20% to 10%, priced each extra editor seat at $50 a month and raised the cheapest Business plan from $125 to $150.

The same Pro and Business prices are still in the code of today’s pricing page, in a calculator the page doesn’t show, so they may not be far off. The pricing data that its credits table loads, which that page doesn’t display, has the same prices with more credits on Pro, 12,000, 18,000 or 24,000 a month. But Voiceflow’s current docs show three changes since. Free no longer renews each month. On self-serve plans, the credits table now gives every rate in dollars. And where some reviews say you can’t buy more credits mid-month, today you can move to a bigger credit bundle or turn on auto top-ups. Check the price under Plans and Billing before you budget.

What Voiceflow’s credits pay for in a chat

Credits pay for everything your agent does. On self-serve plans, Voiceflow’s credits table prices a chat like this.

• Each message a visitor sends costs $0.005, and so does each request to the Conversations API. The launch event that opens a chat is free.
• The model is charged per 1,000 tokens it reads and per 1,000 it writes, at each model’s own rate. GPT-4.1 mini costs $0.0005 to read and $0.002 to write, GPT-5.5 costs $0.00625 and $0.0375, and Voiceflow’s own Flash 4.2 model costs $0.0003 and $0.001.
• Faster processing, which Voiceflow’s glossary calls priority processing that “uses more credits”, raises GPT-4.1 mini’s rates by 75%.
• Messages from the no-reply feature are free, but a generated one pays for its model tokens.

The same table’s Enterprise tab counts in credits instead, one for each visitor message and ten for a phone minute. So every self-serve rate works out at half a cent a credit.

Some credits go before a customer writes anything. Tests spend credits when they use the model, “the same as in a real conversation”, and so do the evaluations that score your transcripts. A knowledge base source split with an LLM chunking strategy spends credits every time it syncs, so a page set to refresh daily costs something every day. The knowledge base hands the model 3 chunks of your content for each question by default, and Voiceflow’s docs say raising that adds token usage.

Voice is priced on the same table, and it’s what many of this search’s results cover. A phone call costs $0.05 a minute, with speech-to-text from $0.0024 a minute and text-to-speech from $0.000625 per 1,000 characters on top, depending on the vendor you choose. The rest of this page is about chat.

A month of website chat on Voiceflow, priced

You can’t price a plan without signing up, but you can price the usage. Here’s a month for a small SaaS site, worked out from the rates above.

The site has 500 chats a month, and each visitor sends 2 messages, close to the 1.95 answers a conversation Deacon measured across 1,053 conversations on 22 September 2026. That’s 1,000 visitor messages and 1,000 replies. We assume the model reads 3,000 tokens for each reply, covering your instructions, 3 chunks of your docs and the conversation so far, and writes 200. That’s our assumption, not Voiceflow’s, and an agent that calls a tool or rewrites the question before it answers reads more.

Visitor messages are $5 of each figure at 500 chats and $20 at 2,000, and the rest is the model. The model’s share on GPT-5.5 is about 14 times what it is on GPT-4.1 mini, so the model you pick matters more than how many people write in. At half a cent a credit, the 500-chat month on GPT-4.1 mini is 1,380 credits. Pro’s smallest bundle came with 10,000 a month in April 2025, and 12,000 in the pricing data behind today’s credits table, though no Voiceflow page prints today’s figure.

So for chat at a founder’s scale, the usage is small. It comes out of your plan’s monthly credits, so unless you run past them, your bill is the plan fee plus any editor seats you add, and that’s the figure Voiceflow shows only inside the product.

What happens when Voiceflow’s credits run out

When credits run low, Voiceflow’s billing docs offer a higher plan, a bigger credit bundle or auto top-ups, and none of them happens unless you choose it.

• A bigger credit bundle is under Plans and Billing, then Billing, Manage and Edit credit bundle.
• Auto top-ups stay off until you enable them. Voiceflow then buys a dollar amount you set each time your balance hits zero, and nothing is charged in advance.

Without either, Voiceflow says what happens on the phone but not in chat. Since 11 August 2026, its changelog says, a call to an organisation that’s out of credits is declined, the caller hears that no agents are available, and the organisation’s admins get an email. We found no page saying what a chat visitor sees. Voiceflow’s glossary files auto top-ups under questions such as “stop the bot from going down when credits run out”.

One setting caps what a single chat can spend. Since 30 July 2026, a cost limit per conversation ends a session once its cost passes a dollar amount you set, with an optional closing message. It’s off by default, and we found no cap on a month’s total spending in Voiceflow’s docs. Your plan’s credits reset at the start of each billing cycle.

What a chat agent on Voiceflow needs besides credits

Three things add to what chat on Voiceflow costs you, in money or in time.

• You build the agent. It’s made from playbooks, which let the model work towards a goal, and workflows, which you lay out step by step, and you publish a version before the chat widget can go on your site.
• A person takes over through another product. Voiceflow’s live agent handoff passes a chat to Genesys, Ujet, Kustomer, Dixa or Zendesk, so you need one of those running. On Zendesk, each handoff opens a messaging ticket that your team answers from Zendesk’s Agent Workspace.
• The widget shows Voiceflow’s branding, and its widget docs say customers on a Business plan can turn it off.

What Deacon costs for website chat

Deacon is an AI customer support agent for founders who handle support themselves, and it lives in a chat widget on your site. There’s nothing to design and no version to publish. Add one script tag, or hand the prompt on the Install page to a coding agent such as Claude Code, Cursor or Codex, then point Deacon at your help pages, files or pasted text. It answers in seconds, at any hour, in the language each visitor writes in.

The prices are on the pricing page, and you can read them without an account. The free plan gives you 50 answers a month, one seat and about 67 pages of docs, with every feature, no card and no end date. Starter is $50 a month, or $40 a month billed yearly, for 1,000 answers and 3 seats. Growth is $150 a month, or $120 a month billed yearly, for 4,000 answers and 10 seats. A conversation takes about 2 answers, so that’s about 25, 500 and 2,000 conversations a month.

There’s one meter. An answer is one reply Deacon writes on your live site, and every plan answers with the same model, so there are no token rates and no model to choose. Chats in the Playground and from localhost are never counted. Nothing is billed past your plan’s answers, and no plan upgrades itself. Your team’s owners get an email when 80% of the month’s answers are used, and again at 100%. After that, visitors get a short reply saying Deacon can’t look into it right now and are asked for their email, unless they’ve already left it. There’s no “powered by” line on any plan, so there’s no badge to pay to remove.

When your docs don’t cover a question, Deacon says it doesn’t know instead of guessing, and asks for the visitor’s email. You get an email when they leave one or ask for a person, and you reply from Deacon’s own dashboard, with no contact centre to connect. Your reply appears in their open chat under your name, and Deacon stops answering in that conversation until you hand it back. Messages between you and the visitor never count as answers.

Each question your docs didn’t cover waits in the dashboard with an Answer this button. The answer you write becomes part of what Deacon knows, and Save & check shows you the reply it now gives and how many other open questions your answer covers. So Deacon gets better the more you answer, and needs you less.

It also shows you what customers need. Every conversation is saved and searchable by what visitors asked. Topics group their questions, count how many people asked and label most of them, as a how-to, a feature ask or a bug report among others. The same script runs your website analytics, with visitors, sources, goals and funnels, so there’s no second tag to add.

Which one fits you

If you need a phone agent, or you want to design every step of an agent and hand chats to a contact centre, Voiceflow is built for that, and its rates are clear once you’re inside. If you answer support yourself and want answers on your website this week, Deacon is the one built for you. It answers your customers from your docs in seconds, hands you only the questions that need you, and shows you what they’re stuck on and what they want next. Every feature is on the free plan, and it needs no card.

There’s more in Voiceflow alternatives, Botpress vs Voiceflow and Chatbase vs Voiceflow. Botpress pricing prices the platform closest to Voiceflow, and how much an AI chatbot costs prices one month on eleven tools.

Questions about Voiceflow pricing

Is Voiceflow free?

For trying it out. Voiceflow’s Free plan comes with a one-time grant of credits that doesn’t renew, and its billing docs say that makes it unsuitable for production. Agencies and partners get a free trial with no card from the pricing page. If you cancel a paid plan, you keep it to the end of the period and then drop to the free tier.

How much is Voiceflow Pro?

Voiceflow shows the current price only inside the product, under Plans and Billing. In April 2025, Pro was $60, $90 or $120 a month for 10,000, 15,000 or 20,000 credits, or $648 a year for the smallest bundle, and those prices are still in its pricing page’s code. Its free plan and its credits table have changed since, so check the price inside Voiceflow before you budget.

Does Voiceflow charge per seat?

For editors beyond the ones your plan includes. Every paid plan includes some editor seats, and more are an add-on. In April 2025 an extra editor cost $50 a month. Viewers are free on every plan.

Can I cap what Voiceflow spends?

Partly. Auto top-ups are off until you switch them on, and you set the amount each one buys. A cost limit per conversation, also off by default, ends any session that spends more than a dollar amount you choose. We found no cap on a month’s total spending in Voiceflow’s docs.

Does Deacon take phone calls?

No. Deacon answers in a chat widget on your website, and you reply from its dashboard. It has no voice, phone or SMS channel, and it takes no actions in other systems, such as looking up an order. If you need calls, Voiceflow takes them in and out.

Can I run Deacon’s free plan on my live site?

Yes. The free plan answers on your live site, with 50 answers a month, every feature and one seat, no card and no end date, and nothing on it is deleted if you go quiet. Chats in the Playground and from localhost don’t count towards the 50.