WhatsApp Cloud API: Complete Guide to the Official Meta API

The WhatsApp Cloud API is the official and free hosting method that Meta offers for businesses to chat with customers at scale within WhatsApp. If you've ever gotten lost among acronyms like BSP, template, 24-hour window, and per-message pricing, this guide demystifies everything: what it is, how it works, what it really costs, and when it makes sense for your business.

TL;DR

The WhatsApp Cloud API is hosted by Meta itself, with no server cost — you only pay for the messages you send.

Since July 1, 2025, billing is per delivered message, not per conversation.

Customer-initiated conversations (service) are free and unlimited within the 24-hour window.

Sending limits start low (250 messages/day without verification) and increase based on verification, quality, and volume.

The API is just the plumbing: you still need a service platform on top of it.

What is the WhatsApp Cloud API (and what it is not)

The WhatsApp Cloud API is one of three ways a business can connect to WhatsApp within the so-called WhatsApp Business Platform — Meta's set of corporate products. In practice, it's an application programming interface (API) that your company, or the software you hire, uses to send and receive messages automatically, without anyone typing on a phone.

The key word is Cloud: unlike the old On-Premises version, here Meta itself hosts all the infrastructure on its servers for free. You don't maintain a server, don't update versions, don't worry about uptime of the plumbing. This drastically lowers the technical barrier to get started.

Now, what it is not: the Cloud API is not an application. It has no screen, no inbox, no send button. It's an engine. For a human to serve, or for an AI agent to respond, you need a software layer on top — whether a ready-made platform or your own system. Confusing the API with the complete solution is the number one mistake for beginners.

Cloud API, On-Premises, and Business App: the three entry points

There are three official ways to use WhatsApp professionally, and choosing the wrong one costs time and money. The table below summarizes the differences.

Feature	WhatsApp Business App	Cloud API	On-Premises API
Who hosts	Your phone	Meta (cloud)	Your server
Infrastructure cost	Free	Free	Server + maintenance
Simultaneous agents	Up to 5 devices	Unlimited	Unlimited
Real automation	Limited	Full	Full
Status in 2026	Active	Recommended	Discontinued

The Business App is the free app from the store, ideal for micro-businesses with low volume — but it stalls when you need multiple agents and serious automation. The On-Premises API was the old version that ran on your server; Meta discontinued it and ended support in October 2025, migrating everyone to the cloud. So today, the WhatsApp Cloud API is the default choice for any operation that takes service seriously. If you're still torn between app and API, it's worth reading our detailed comparison at WhatsApp Business App vs Official API.

How the WhatsApp Cloud API works internally

Understanding three concepts solves 90% of the confusion about the official API. They are templates, the 24-hour window, and webhooks.

Message templates

You cannot just send any text to someone who hasn't contacted you first. To start a conversation, the business must use a template (message model) previously approved by Meta. Each template has a category — marketing, utility, or authentication — and this category determines how much you pay. Templates are precisely what protects users from spam and maintains the quality of the ecosystem.

The 24-hour service window

When a customer sends you a message, a 24-hour service window opens. Within it, you can reply freely with any content, without a template and — since November 2024 — without paying anything. After 24 hours without the customer replying? To reopen the conversation, you need a paid template. This mechanism is the heart of the cost model.

Webhooks

The Cloud API communicates with your system via webhooks: every time a message arrives, Meta fires a notification to a URL you configure. That's how your platform knows the customer replied and can trigger a human or an AI agent in real time.

How much it costs: the per-message pricing model

Here lies the biggest source of myths. Until mid-2025, Meta charged per conversation (24-hour blocks). Since July 1, 2025, the model changed to per delivered message (per-message pricing): you pay for each template message that reaches the recipient, and the amount depends on the category and the country of the receiving number.

Category	When it is used	Billing
Service	Reply within the 24-hour window	Free
Utility	Confirmations, receipts, order updates	Paid (free within the window)
Authentication	Verification codes (OTP)	Paid, per message
Marketing	Promotions and campaigns	Paid, usually the most expensive

Two points that almost no one explains properly: first, service messages are free and unlimited — if your business is mostly reactive (customer calls, you answer), messaging costs can be near zero. Second, Meta now offers lower rates by volume for utility and authentication templates. Rates vary greatly by country, so always check the official table for the category you use most.

To make this concrete, think of a store that sends an order confirmation (utility) and the customer replies with a question. The confirmation within the open window may be free, and all subsequent service exchange as well. On the other hand, a promotional campaign sent to a thousand contacts who haven't interacted recently falls into the marketing category — the most expensive — and each delivered message is charged. The practical lesson is straightforward: the more your flow is service and utility, the lower the cost; the more it is cold marketing, the higher. Designing flows with this in mind is what separates a lean operation from an expensive one.

Watch out for a hidden cost: the intermediary. Many platforms add a markup on top of Meta's price per message. In volume, this becomes a significant amount. We detail this trap in Markup on WhatsApp Messages.

Sending limits and throughput: how to scale without being blocked

The WhatsApp Cloud API does not release unlimited sending on day one — and that's intentional, to curb spam. Every new account starts at a low messaging tier and rises by merit.

Without business verification: 250 messages initiated per business in 24 hours.
Verified and with good quality: limits increase in steps — 1,000, 10,000, 100,000, and in some cases, unlimited.
Tier advancement: since October 2025, the limit is counted per Business Portfolio (not per number), and Meta reassesses upgrades every few hours.

There is also throughput, which is the sending speed. By default, the Cloud API processes about 80 messages per second, and high-tier accounts can reach 1,000 per second. For most SMEs, this ceiling is never a problem.

What brings down an account is not volume per se, but quality. If many people block or report your messages, your quality score drops, the tier stops rising, and the number may be restricted. Since the official API itself reduces — but does not eliminate — this risk, it's worth understanding best practices in How to Avoid WhatsApp Business Number Blocking.

How to get started with the WhatsApp Cloud API: step by step

Technically, you can set up a test number in minutes. For production, the path is as follows:

Meta Business Manager account — the central dashboard where everything is managed.
Business Verification — submission of company documents; this unlocks higher limits.
WhatsApp Business Account (WABA) — created within Business Manager.
Number registration — a phone number that is not linked to a regular WhatsApp or Business App. This point catches many people.
Template approval — register and wait for Meta's review.
Webhook configuration — to receive responses in your system.

When NOT to do this alone: if you don't have a development team, integrating the raw API is a lot of work and ongoing maintenance. In such cases, hiring a platform that already delivers everything ready on top of the Cloud API is often cheaper than maintaining your own system.

Common mistakes that bring down accounts on the Cloud API

After implementing the WhatsApp Cloud API in dozens of operations, the pitfalls repeat:

Using a personal number already active on regular WhatsApp. A number registered on the API cannot exist simultaneously on the normal app. Migrating requires deleting the old account from that number.
Treating marketing templates as service. Sending promotions to those who didn't ask for them destroys your quality score and cost.
Ignoring the 24-hour window. Trying to reply outside it without a template simply won't deliver.
Buying the API thinking it's the final solution. Without a service layer on top, you have an engine without a car.
Neglecting the template category. Registering something clearly marketing as utility leads to rejection or reclassification by Meta, and can make what seemed cheap expensive.

The good news is that all these mistakes are avoidable with planning. Most of them, in fact, disappear when you use a platform that abstracts registration, categorization, and respect for the 24-hour window. If your number has ever been restricted, migrating to the official API is usually the safest way out — a topic we delve into in WhatsApp Blocked: The Official API is the Solution.

The Cloud API is the engine, not the car

It's worth repeating, because it's the most expensive misunderstanding: the WhatsApp Cloud API alone doesn't serve anyone. It transports messages. What organizes queues, distributes conversations among agents, stores history, triggers automations, and puts an AI agent to respond 24/7 is the platform that runs on top of the API.

That's exactly the role of solutions like Voyia, which use the official API underneath and deliver the complete service experience — without you having to deal with webhooks, templates, and Business Manager by hand. If your team currently pays per agent, it's worth seeing why that model no longer adds up in Unlimited Agents on WhatsApp.

Conclusion: when the WhatsApp Cloud API makes sense

If your business needs multiple agents, real automation, and wants to be on the right side of Meta's rules, the WhatsApp Cloud API is the way — free to host, pay only per message, and increasingly with free service. The secret is to stop seeing it as a final product and treat it as the foundation on which you (or a partner) build service.

The next step is simple: list today how many agents and what message volume you actually have. With those two numbers in hand, it becomes easy to decide between the Business App, the raw API, or a ready-made platform — and stop paying for what you don't use.