A static bot follows a fixed flow that you design: menus, buttons and set replies, with no AI involved. An AI bot uses a language model you choose to understand each message and write its own reply, and can answer from your documents or call your APIs. Meta Business Agent is Meta’s own AI agent, run and billed by Meta.

A flat amount for each conversation the bot handles, however many steps the flow has.

Every AI reply uses tokens: the text the bot reads (your instructions, the conversation so far and any documents it looks up) and the text it writes back. The cost of a reply depends on the model you choose and how many tokens it uses.

Meta charges one rate of $2.00 per 1 million tokens, which covers both the AI processing and delivering the message. Meta says a reply typically uses 20,000 to 25,000 tokens, so roughly 4 to 5 US cents a reply. These messages are always charged, even in Meta’s free windows.

Lighter models such as Gemini 3.1 Flash Lite or GPT-5 Mini cost a fraction of the larger ones and handle most FAQ and support conversations well. Larger models are worth it when the bot has to reason over long documents or make several tool calls.

No. This calculator covers the bot itself. Messages your bot sends are also WhatsApp messages, charged at the per-message rates on the pricing page.