WhatsApp Business app vs API
This page compares the WhatsApp Business app, Coexistence, and a dedicated Cloud API number on the things that actually differ: who can reply, what happens to history and groups, how many messages you may send, and what it costs. The rules quoted here are Meta’s own, with the dates they change.
- Three ways to run one number, compared on devices, history, groups, limits and cost
- The 24 hour customer service window, and what it forces you to do differently
- Messaging limit tiers, and the two ways to move up them
- What changes on 1 October 2026, when in-window replies stop being free
Short answer
The app and the Cloud API are not two products you choose between once and forever. There are three ways to run one business number: the WhatsApp Business app on its own, the app and the Cloud API on the same number at the same time (Coexistence), or a dedicated Cloud API number with no app on it.
Most teams that outgrow the app take the middle option, because it adds software without giving up the phone, the number, or the way people already work.
The three ways to run one number
The WhatsApp Business app is the free phone app. One phone anchors the account and up to four linked devices work alongside it. A Meta Verified or Meta One subscription raises that to ten devices and adds chat assignment between named agents inside the app itself.
Coexistence keeps that app and connects the same number to the Cloud API as well. Messages are mirrored between the two, so a team inbox and the app both work on the number at once. This is the path for a business that wants software without a migration.
A dedicated Cloud API number has no app on it. Every message goes through the API, which means everything can be routed, owned, measured and automated, and nothing can quietly bypass the workflow by someone picking up the phone.
Those are not stages of maturity. A five person team can run happily on the app for years, and a two person team selling high-value contracts may want a dedicated API number on day one.
What each one can and cannot do
Every cell below is either a limit Meta publishes or a direct consequence of one. Where the official documentation does not answer a question, the cell says so instead of guessing.
| WhatsApp Business app | Coexistence | Dedicated Cloud API | |
|---|---|---|---|
| People who can reply | One phone plus up to four linked devices, or ten with a Meta Verified or Meta One subscription | Everyone in the app, plus everyone in the connected inbox | Everyone in the connected inbox |
| Who owns a conversation | Nobody, unless you subscribe: multi-agent adds chat assignment and shows which agent and device sent a message | The inbox owns it. The app has no owner field | The inbox owns it |
| Chat history | On the phone | The most recent 6 months of chat messages sync in. Group chat messages are not included | Starts at connection. There is no history to sync |
| Groups | Yes | Group chat messages are not included in the sync | Not part of the Cloud API messaging surface |
| Live location, broadcast lists, status | On the primary phone. Not supported on linked devices | Same as the app for the app side | Not part of the Cloud API messaging surface |
| Calls | All devices when you have four or fewer linked. Primary device only above four | Same as the app for the app side | Handled separately from messaging |
| Throughput | Not published as a rate | Fixed at 20 messages per second | Governed by your messaging limit tier |
| Cost | Free | App-sent messages stay free. Cloud API messages are billed at Cloud API pricing | Cloud API pricing |
| Removing someone | Unlink their device from the primary phone | Both: unlink the device and deactivate the inbox account | Deactivate the inbox account |
Device counts, sync scope and the 20 mps ceiling are published by WhatsApp and Meta and are cited at the bottom of this page. Ownership and removal describe how a connected team inbox works, using marv.Inbox as the reference.
The 24 hour window, and why it changes how you work
On the Cloud API you cannot message whoever you like whenever you like. A message from a user opens a 24 hour customer service window. Inside that window you can send ordinary messages of any type. Outside it, template messages are the only type you can send, and templates have to be created and approved in advance.
Messages a user sends to you are never charged. The window closes 24 hours after that user last messaged you.
There is one exception worth knowing. If a user reaches you from a click-to-WhatsApp ad or a Page call-to-action button and you reply within 24 hours, a free entry point window opens for 72 hours, and while it is open every message you send is free regardless of type. The customer service window runs independently, so once that 24 hours is up you are back to templates only even if the free window is still open.
The practical effect is that follow-up has to be designed rather than improvised. A conversation you leave for two days cannot be picked up with a casual message. That constraint does not exist in the app, and it is the single biggest working difference between the two.
How many people you may message
Cloud API numbers carry a messaging limit, which caps how many unique users you can start conversations with. The tiers are 250 for a newly created business portfolio, then 2,000, 10,000, 100,000 and unlimited.
There are two ways to reach 2,000: verify your business, either directly or through a solution provider, or deliver 2,000 messages outside customer service windows to unique user phone numbers within a rolling 30 day period, using templates with a high quality rating.
Above 2,000 the scaling is automatic. If you are sending high-quality messages across all of your business phone numbers and templates, and you have used at least half of your current limit in the last 7 days, your limit rises one level within 6 hours.
Read that as a quality mechanism rather than a volume one. The limit is not a plan you buy, it is a rating you earn, and templates that people block or report will hold you where you are.
Connecting, and changing your mind later
Coexistence and a full Cloud API connection are two different Meta login flows, not one flow with a checkbox. In our own implementation they are separate configured flows chosen before the dialog opens, which is worth knowing because it means the decision is made at connect time rather than adjusted afterwards.
On the Coexistence path there is a real deadline in the middle of it. Meta gives the connecting partner 24 hours to synchronize messaging history, and a connection that does not complete that has to be offboarded and started again.
Coexistence is also the only one of the three that documents a history sync. A dedicated Cloud API number starts from nothing, because there is no app on it to sync from. If keeping context matters more than a clean separation, that is the argument for Coexistence in one sentence.
What Meta’s documentation does not spell out is exactly what happens to an app that is migrated to a dedicated Cloud API number without Coexistence. We are not going to fill that gap with a confident sentence. If it matters to your decision, ask before you connect, not after.
Which route fits you
| If this is you | Route | The thing that decides it |
|---|---|---|
| Solo, or two people who never share a customer | WhatsApp Business app | There is no coordination problem to solve, so software adds cost and no answer. |
| Up to ten people who need chats divided between them | App with a Meta Verified or Meta One subscription | Assignment inside the app, without a platform migration or a per-message bill. |
| A team sharing one queue that needs history, notes, SLA and reporting | Coexistence with a team inbox | You keep the number, the phone and six months of history, and gain a record of who did what. |
| Support or sales run as an operation, with routing and accountability | Dedicated Cloud API number | Nothing bypasses the workflow, because there is no app to bypass it with. |
| High-volume outbound campaigns | Dedicated Cloud API number | Coexistence is capped at 20 messages per second, which is a real ceiling for bulk sending. |
The tradeoffs nobody advertises
Moving a number onto the Cloud API buys control and costs flexibility. These are the specific costs.
- ▲Outside the 24 hour window you can only send approved templates, so re-opening a cold conversation is a process rather than a message.
- ▲A new business portfolio starts at 250 unique users. That is a real constraint on a launch, and verification is the fastest way past it.
- ▲Coexistence is fixed at 20 messages per second, however large the plan around it.
- ▲Coexistence syncs the most recent 6 months of chat messages and excludes group chat messages. It is not an archive migration.
- ▲Meta Business Agent messages have been charged per token since 1 August 2026, at $2.00 per 1M tokens. They used to be free.
- ▲From 1 October 2026, service messages and utility messages sent in response to users become chargeable. Meta’s wording is that any non-template message is charged as of that date. Any cost model you build on today’s free in-window replies has a known expiry.
- ▲Messaging limits are earned through template quality, not purchased. A poor quality rating holds a number at its current tier.
Questions we get asked
Do I have to move off the WhatsApp Business app to use an inbox?+
No. Coexistence connects the same number to the Cloud API while the app keeps working, with messages mirrored between them. A dedicated Cloud API number is the alternative when you specifically want nothing bypassing the workflow.
Why can I not just reply to an old conversation on the API?+
Because non-template messages can only be sent inside an open 24 hour customer service window, and that window closes 24 hours after the customer last messaged you. Outside it, template messages are the only type you can send.
How do I get my messaging limit raised?+
Verify your business, or deliver 2,000 messages outside customer service windows to unique numbers within a rolling 30 days using high quality templates. Above 2,000 it scales automatically when you use half your limit in 7 days and keep quality high.
Is the API more expensive than the app?+
Today the app is free and the Cloud API charges per delivered template, with non-template in-window messages free. From 1 October 2026 that second half changes: any non-template message is charged. Model both dates before you commit.
Can I keep my groups?+
Group chats stay in the app. Meta states that group chat messages are not included in the Coexistence sync, and groups are not part of the Cloud API messaging surface.
Sources, checked June 2026
Sources
Device limits, sync scope, throughput, window rules, messaging tiers and the dated pricing changes all come from Meta and WhatsApp official documentation. Meta changes these without notice, so the checked date matters as much as the link.
- WhatsApp Help Center: About linked devices on the WhatsApp Business app
- WhatsApp Help Center: About multi-agent
- Meta for Developers: Onboard WhatsApp Business app users (Coexistence)
- Meta for Developers: Pricing on the WhatsApp Business Platform
- Meta for Developers: Upcoming pricing updates for non-template messages
- Meta for Developers: Messaging limits
Checked on 2026-08-03
Who wrote this
marv.Inbox team
Product and support team, marv.Inbox
We build marv.Inbox and we run our own customer conversations inside it, on the same WhatsApp number the buttons on this page open.
Compare it live, not on a feature grid
Open the interactive demo and judge ownership, SLA timers, and handoff for yourself before you shortlist.