Meta confirmed a fundamental change to WhatsApp Business API’s pricing model, rolling out in stages through the second half of 2026. The short version: starting August 1<\/strong>, Meta begins charging for messages processed by its new Meta Business Agent (the AI agent it launched on July 1), and starting October 1<\/strong>, service and utility messages inside the conversation window \u2014 free until now \u2014 also become paid. If your business uses WhatsApp to support customers, this hits your budget directly, not just your operations.<\/p>\n\n\n\n Until now, the model was simple: marketing and utility messages that opened a conversation were charged per template, but once a customer wrote first, the business could reply for free within the 24-hour window. That \u00abfree\u00bb is exactly what Meta is removing in October. For a B2B SMB handling support, scheduling, or order confirmations over WhatsApp, that free window covered the bulk of daily message volume.<\/p>\n\n\n\n If your company uses WhatsApp Business API to confirm orders, answer support questions, or coordinate deliveries, most of that back-and-forth happens inside the 24-hour window that’s free today. Starting in October, every one of those replies will carry a cost. The more support volume you run through WhatsApp, the bigger the hit to your monthly bill. This matters even more if you’ve already adapted your chatbot to the specialized-agent rules Meta rolled out in July<\/a>: that same bot is now about to start costing money per reply, too.<\/p>\n\n\n\n The other important change is the per-token charge for Meta’s own AI agent. If your business decides to use the Meta Business Agent instead of building its own flow, cost no longer depends only on conversation volume \u2014 it depends on how much processing each reply requires. A simple question (\u00abdo you have this in stock?\u00bb) costs less than one that requires the AI to check order history or pull data from multiple sources.<\/p>\n\n\n\n The most direct way to contain the impact is to segment conversations by complexity. Simple, repetitive questions \u2014 hours, order tracking, contact details \u2014 can be handled with low-cost automated replies, while you reserve more sophisticated (and pricier) AI support for cases that actually justify it. If you’ve already looked at other AI customer-support options like ChatGPT Live<\/a>, now’s the time to compare real cost per conversation across channels, not just their capabilities.<\/p>\n\n\n\nThe exact timeline of changes<\/h3>\n\n\n\n
\n
What this means for a B2B SMB<\/h2>\n\n\n\n
How to prepare before August 1<\/h2>\n\n\n\n
\n
Cut costs without losing service quality<\/h2>\n\n\n\n