20 July 2026 · WhatsApp
WhatsApp Business API pricing in Singapore, explained.
Accurate as of July 2026
Meta's WhatsApp Business pricing changes again on 1 August 2026 and 1 October 2026. This article describes the pricing model and the direction of change. For exact per-message figures, check Meta's current rate card for Singapore.
If you run a clinic or service business in Singapore and you are weighing a WhatsApp bot, the pricing model matters more than any single number, because that model is moving twice more this year. This article explains how WhatsApp Business API pricing actually works, why Singapore now has its own rate card, and what the 2026 changes mean for the budget of a WhatsApp-first intake or support system.
You pay per delivered template message, not per message sent.
Since 1 July 2025, WhatsApp Business API billing runs on a per-delivered-template model. A business pays when a template message is delivered, not merely when it is sent. The practical effect is that you are not charged for messages that never reach the customer, but you are charged for every one that does, including the ones inside an ongoing conversation that fall outside the free-form windows.
Conversations on the WhatsApp Business API are built around categories: broadly, marketing, utility, authentication, and service. Each category is priced differently, and which category a message falls into determines what you pay. A reminder or appointment confirmation is typically a utility message; a promotional broadcast is marketing; a one-to-one reply inside a customer-initiated window has historically been treated as a service conversation. The boundaries between these, and what is chargeable within them, are exactly what is shifting in 2026.
Singapore now has its own rate card.
Singapore is no longer folded into a broader regional rate card; it has a standalone one. That matters because it means Singapore pricing can move independently of neighbouring markets, and the specific categories a Singapore business uses most (utility reminders, authentication, service replies) are priced on their own line. For a local clinic sending appointment reminders and intake replies, the relevant figures are the Singapore utility, authentication, and service rates, not a regional average.
Three changes landing in 2026.
The model is not settling. Three further changes are scheduled, and each nudges the cost of running a WhatsApp system in Singapore:
- From 1 July 2026: utility and authentication rate increase. The categories a clinic leans on hardest (utility reminders, authentication) move upward. This is already in effect.
- From 1 August 2026: per-token billing for AI replies. AI-generated replies move to per-token billing. For a bot that holds a real conversation, this ties a slice of cost to how much text the model generates, a different shape of cost from a fixed per-message fee.
- From 1 October 2026: service and in-window utility messages become chargeable. Messages that sit inside a service conversation, and in-window utility messages, become chargeable. This is the change most likely to affect a support-heavy or intake-heavy bot, because the back-and-forth of a real conversation starts to accrue cost where it previously did not.
What this means for a Singapore clinic or service business.
The honest takeaway is that a WhatsApp system is not a fixed monthly line item. It is usage-priced, and the usage a busy clinic generates (reminders, intake replies, follow-ups) is exactly the usage moving through these changes. Budgeting for it means modelling the categories you send most, not grabbing a single number.
Two practical implications follow. First, the October change makes it worth being deliberate about how much back-and-forth a bot does inside a service window. Every reply has a cost attached, so a bot that qualifies and routes efficiently is cheaper to run than one that chats at length. Second, the August move to per-token AI billing means the cost of AI-generated replies will scale with verbosity, which is a reason to keep automated replies tight and useful rather than expansive.
None of this is a reason to avoid WhatsApp. For a Singapore clinic, it is still the channel patients actually use. It is a reason to size the system around the real usage and to check the current rate card before committing to a number.
Where to check the exact figures.
Because the rates move on a published schedule and Singapore has its own card, the only reliable source for exact per-message figures is Meta's current WhatsApp Business pricing rate card for Singapore. Pull it when you budget, and pull it again after 1 October 2026 once the service- and utility-window changes have landed. Anything quoted here would be out of date within months; the model and direction above are what stay stable.
Questions we get about WhatsApp pricing
- How much does the WhatsApp Business API cost in Singapore?
- There is no single figure. Billing is usage-based: you pay per delivered template message, and the rate depends on the message category: marketing, utility, authentication or service. Singapore has its own standalone rate card rather than sitting inside a regional average, so the only reliable source for exact per-message figures is Meta's current rate card for Singapore.
- Is WhatsApp Business API a fixed monthly fee?
- No. It is usage-priced, which is why a busy clinic and a quiet one running the same system pay different amounts. Budgeting for it means estimating the volume of the categories you actually send, usually appointment reminders and intake replies, rather than taking one headline number.
- Do I get charged for replies inside a conversation a customer started?
- From 1 October 2026, messages inside a service conversation and in-window utility messages become chargeable. That is the change most likely to affect an intake-heavy or support-heavy bot, because the back-and-forth of a real conversation starts to accrue cost where it previously did not.
- What is the difference between utility, marketing and service messages?
- The category decides the price. An appointment reminder or confirmation is typically a utility message. A promotional broadcast is marketing. A one-to-one reply inside a window the customer opened has historically been treated as a service conversation. Authentication covers one-time codes. Which category a message falls into is what determines what you pay for it.
- Does an AI-written reply cost more than a normal one?
- From 1 August 2026, AI-generated replies move to per-token billing, so a slice of the cost scales with how much text the model produces. Practically, that is a reason to keep automated replies tight and useful rather than expansive. A bot that answers and routes efficiently is cheaper to run than one that chats at length.
Planning a WhatsApp bot for your clinic?
We build WhatsApp intake and booking systems for Singapore clinics, sized to real usage, usually live in one to three weeks.