Meta Business Agent billing starts 1 August at $2 per million tokens

Meta begins charging for its Business Agent on 1 August 2026 at $2.00 per million tokens, ending the one-month free trial that opened when the platform launched globally on 1 July. A typical interaction consumes 20,000 to 25,000 tokens, translating to roughly 4 to 5 cents per message.

For the more than one million businesses that deployed the agent in the six weeks since launch, this is the first time AI-powered customer service on Meta's platforms will carry a direct cost.

The pricing applies to all AI-handled conversations across WhatsApp, Instagram, and Messenger, and marks a meaningful shift in how businesses will need to budget for automated customer engagement at scale.

Two pricing waves in three months

The 1 August billing date is the first of three pricing changes Meta is rolling out on the WhatsApp Business Platform in the second half of 2026.

The second wave arrives on 1 October, when Meta resumes charging for service messages, the free-form replies that businesses or their human agents send inside the 24-hour customer service window, which have been free since November 2024.

From October 1, there is no free way to respond to a customer inside an open conversation, regardless of whether the reply comes from the AI agent, a human, or a template. Every response carries a cost.

The sequencing matters. The agent becomes billable two months before the broader service-window charges resume, so businesses get a period where the agent is metered but human and template responses inside the window are still free. That interim window is a natural time to shift volume onto the agent and measure the trade-offs before the October changes make every response type billable.

What changes on Monday for businesses

Meta has confirmed that billing begins 1 August, with monthly invoices issued at the end of each billing period. For businesses that have already connected the agent to their catalogs and systems, that transition happens automatically.

The billing structure combines two costs that previously came separately. Meta is bundling the AI processing charge and the message delivery fee into one token-based rate. That means businesses no longer pay for AI usage and message delivery as separate line items. One meter covers both.

The cost model is usage-based, not flat-rate. Token-based billing is fine until you run a campaign that works. A post that brings in thousands of conversations is a great day for your business and a more expensive day for your invoice, and you will not know the final number until the month closes.

At 4 to 5 cents per typical conversation, the immediate cost for many businesses will be manageable. A business handling 10,000 customer interactions per month would spend $400 to $500 on Meta Business Agent billing. The challenge is that complex, multi-turn conversations consume significantly more tokens, and final costs scale with both volume and interaction depth.

Why this matters beyond the headline rate

Meta launching its own AI agent for customer service was always going to reshape the market. Making it free for a month drove adoption. Putting a price on it forces budget conversations that most social teams have not had to have yet.

The meta shift here is not really about the $2.00 per million token rate. It is that Meta has now committed to a commercial model for AI-powered business messaging. The free window was always temporary, and Meta telegraphed that this was a testing period from the start. What 1 August represents is the moment the Meta Business Agent moves from an experiment into a real line item on a real budget.

The token pricing model itself aligns Meta with how AI providers charge for language models, but it introduces variability that businesses have not dealt with on WhatsApp before. A standard customer interaction typically consumes between 20,000 and 25,000 tokens, which translates to roughly $0.04 to $0.05 per message. Unlike traditional WhatsApp tier pricing, which varies heavily by destination country, the token rate for the Meta Business Agent is universal.

The global rate is a notable simplification, but the trade is predictability. Flat per-message pricing is easier to model. Token consumption varies by conversation complexity, which means the bill for August will teach businesses their actual cost per interaction in a way the free trial could not.

The October deadline is the one to model now

1 August is the immediate date, but the material budget impact for most businesses will land on 1 October 2026, when Meta will begin charging for service messages. In addition, utility messages sent in response to users within the 24-hour customer service window will also become chargeable on a per-message basis.

Starting 1 October 2026, WhatsApp will begin charging for service messages, the free-form replies your team or chatbot sends to customers. These have been free since November 2024, so this is a meaningful shift for anyone using WhatsApp for support conversations.

Service message rates will vary by market and will be aligned with the rates for utility messages and authentication messages. Meta has not yet published the October rate card, but it is committed to publishing them by 1 September 2026.

For businesses running high-volume customer service or third-party AI chatbots inside WhatsApp, October is the date that changes the economics. If you have been relying on the free service window to keep WhatsApp support costs manageable, that window is closing in two months.

What to do between now and October

If your business is already running the Meta Business Agent, the next four weeks are your last chance to measure token consumption at no cost. For businesses that have not yet tested the agent, the remaining days without a bill represent the lowest-cost opportunity to learn their own token consumption patterns before committing to a budget.

The practical steps are straightforward. Audit your current WhatsApp message volume, separate AI-handled conversations from human-handled service messages, and model both the August and October cost impacts before the charges begin.

If your volume is high and your margins are tight, the August-to-September window is the time to test whether shifting more conversations to the Meta Business Agent reduces your overall October bill, or whether the token costs outweigh the benefit.

The combination of the 1 August AI billing launch and the 1 October service message fee restoration amounts to a meaningful change in what it costs to run customer communication through Meta's platforms in the second half of 2026. Neither change is dramatic in isolation, but running both scenarios through your current messaging volume before August is worth the twenty minutes it takes.

Context: how this pricing compares

Meta's $2 per million token rate sits within the broad range of AI customer service pricing, but direct comparisons are complicated by the fact that different platforms measure and charge differently.

Fin is priced at $0.99 per outcome. Zendesk charges $1.50 per automated resolution (committed volume) or $2.00 pay-as-you-go for overages. Gorgias charges approximately $0.60 to $1.27 per automated resolution depending on tier.

The challenge is that a per-outcome price and a per-token price are measuring different things. A single outcome might involve multiple back-and-forth exchanges, each consuming tokens. The Meta Business Agent charges for every token processed, not just for successful resolutions, which means costs accumulate across the entire conversation regardless of whether the issue was resolved.

What Meta's pricing does offer is transparency on the unit cost. You pay for what the AI processes, not for an abstracted bundle. That makes it easier to understand the cost drivers, but harder to predict the monthly total until you have real usage data.

The bigger picture

Over a million businesses were already using an earlier version in pilot markets including India, Mexico, and Brazil when Meta announced the global expansion at its Conversations 2026 conference in London on 3 June 2026.

The product itself is more capable than a basic FAQ bot. The Meta Business Agent can answer product questions against a connected catalog, recommend specific items, book appointments, qualify incoming sales leads, and process transactions, all without a human agent typing a reply. The platform connects to Shopify, Zendesk, and a growing list of integrations.

For businesses that have been waiting to see whether AI customer service would become a standard channel cost or remain experimental, the answer is now clear. It is standard, it is billable, and it is live in two days.