Chatbase pricing explained: plans, credits and what you really pay

We make intoCHAT, an AI chatbot builder that competes with Chatbase. This page is not affiliated with, endorsed by or sponsored by Chatbase; Chatbase is a trademark of its owner. We wrote it because Chatbase's prices are spread across a pricing page, a plan comparison table and several docs pages, and the cost of a plan depends on the AI model you pick.

Every Chatbase number below comes from Chatbase's own pricing page and documentation, checked on 10 October 2026 and linked under Sources at the end. Prices are in US dollars as listed, before any tax. Prices change, so check Chatbase's pricing page before you buy. Where Chatbase does not publish a number, we leave it out rather than guess.

Checked on 10 October 2026 on chatbase.co/pricing. Yearly billing is 20% off the monthly price. Hobby, Standard and Pro show a 7-day trial.

Three add-ons have a published price. Add-ons need a paid plan, and their billing interval must match the plan: monthly add-ons on monthly plans, yearly add-ons on yearly plans.

A message credit is Chatbase's unit of usage. According to its pricing FAQ, each request to the AI model costs a number of credits that depends on the model you choose. Most replies are a single request. With actions enabled, one reply can take several: Chatbase's own example is a booking, where the model first runs a calendar action to fetch free slots and then writes the answer, which makes two requests and twice the credits.

Subscription credits renew on the 1st of each calendar month, whatever day you subscribed, and unused credits don't carry over. Auto-recharge credits don't expire. When credits run out and auto recharge is off, the agent shows a message that it is currently unavailable until the credits renew.

The plan comparison also lists separate monthly test credits: 200 on Hobby, 500 on Standard and 1,000 on Pro. Chatbase's API docs add that voice minutes use message credits; the rate is not on the pricing page.

Chatbase's docs list the credits per reply for every model. A selection:

Chatbase counts credits, not conversations. To estimate your credits, multiply your conversations by the AI replies per conversation, then by the credits per reply of your model. Your own chat history tells you how many replies a conversation takes.

intoCHAT has a monthly message allowance instead of credits. Each message a visitor sends that your agent answers counts as one message with the default model, GPT-6 Luna. Larger models use more per reply. A reply that runs actions or a web search still counts once, at its model's rate.

Removing “Powered by intoCHAT” is included on every paid plan. On monthly Starter and Pro plans you can add agents and team seats for $9 a month each. Starter and Pro come with a 7-day free trial, and the Free plan needs no card.

Usage starts again at the start of each billing period. When the allowance is used up, your agents stop answering until it resets or you upgrade: intoCHAT has no credit top-ups. If fewer messages are left than a model needs, the reply comes from GPT-6 Luna and uses one message instead of being refused.

Several models are offered by both products. On each model below, one reply uses as many intoCHAT messages as Chatbase credits, including Claude Sonnet 5.5 at 2 and Claude Opus 5.5 at 5. What still differs is how many messages or credits each plan includes and what extra volume costs. intoCHAT offers models from OpenAI, Anthropic and Google only, including Gemini Flash (1 message) and Gemini Pro (2 messages). Chatbase also offers models from xAI, DeepSeek, Meta, Mistral and others.

These examples use only the published numbers above. Assumptions: monthly billing, three AI replies per conversation, no extra requests from actions on Chatbase, and the cheapest published combination of plan and add-ons on each side. Your numbers will differ, so redo the sums with your own reply count.

Example 1, a small business website with 200 conversations a month (600 replies) on a 1-credit model. On Chatbase, Hobby's 700 credits cover it: $40 a month, or $139 with the branding removed. On intoCHAT, Starter's 2,500 messages cover it: $19 a month with branding removal included.

Example 2, an online shop with 3,000 conversations a month (9,000 replies) on a 1-credit model. On Chatbase, Standard includes 4,000 credits, and 5,000 more through auto recharge cost $200, so $350 a month; Pro alone would be $500. Removing the branding brings Standard to $449. On intoCHAT, Pro's 10,000 messages cover it: $49 a month. If the agent uses actions, Chatbase may count two or more requests for some replies; intoCHAT counts each reply once.

Example 3, the same shop on Claude Sonnet 5.5, which uses 2 per reply on both. On Chatbase that is 18,000 credits: Pro plus 3,000 auto-recharge credits, $620 a month. On intoCHAT it is 18,000 messages, which is more than Pro's 10,000, and there are no top-ups. You would need an Enterprise plan with custom pricing, or a 1-message model such as GPT-6 Luna, Claude Haiku 4.5 or Auto on Pro at $49. At this volume on a premium model, Chatbase is the self-serve option. At up to 5,000 Sonnet replies a month (about 1,666 conversations), Pro's 10,000 messages cover it for $49, while the same 10,000 credits on Chatbase cost $390: Standard plus 6,000 auto-recharge credits.

Example 4, an agency with 5 client websites and 1,500 replies a month in total on a 1-credit model. On Chatbase: Hobby, 4 extra agents ($100) and 1,000 auto-recharge credits ($40), $180 a month, or $279 with the branding removed. On intoCHAT: Starter with 2 extra agents ($18), $37 a month, or Pro with 10 agents at $49, branding removal included either way.

For a website agent at small to medium volume, intoCHAT costs less in these examples, on a 1-message model and on Claude Sonnet 5.5 up to Pro's 10,000 messages, mainly because its plans include more replies per dollar, include several agents, and include branding removal. Agencies and businesses with several sites benefit most from the included agents.

Chatbase can be the better fit, and sometimes the cheaper one, in other situations: