# Zoom Meeting Bot API | Record & Transcribe from $0.35/hr

> Build a Zoom meeting bot with one API call. Record audio, capture per-participant video, and get real-time transcripts from $0.35/hr. $5 free credit, no infra.

Source: https://meetstream.ai/meeting-bot-api/zoom

## Zoom Meeting Bot API to Record, Transcribe, and Stream Zoom Calls

One POST request sends a named bot into a Zoom meeting and returns the recording, a separate audio and video stream per participant, and a transcript. Works on meetings you do not host, from $0.35 per bot-hour.

### What is a Zoom meeting bot API?

A Zoom meeting bot API lets your application put a participant into a Zoom call programmatically and get the media back. You send a meeting link, a bot joins under a name you choose, and when the call ends you receive an MP4, per-participant audio and video, and a transcript.

The alternative is Zoom's own APIs, which work well when your user is the host on a paid plan and get complicated quickly when they are not. A bot behaves the same way regardless of whose account the meeting sits on, which is why most products that record their customers' meetings use one. This page covers Zoom specifically; the meeting bot API overview covers all three platforms.

### Send a bot into a Zoom call

Create the bot with a single request. The bot joins, records, and fires a webhook when the meeting ends.

```
curl -X POST https://api.meetstream.ai/api/v1/bots/create_bot \ -H "Authorization: Token $MEETSTREAM_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "meeting_link": "https://zoom.us/j/74648392", "bot_name": "Acme Notetaker", "transcription": { "provider": "meetstream" }, "webhook_url": "https://your-app.example.com/meetstream" }'
```

The webhook carries the composed recording plus a separate track for each participant, so you never have to work out who spoke from a single mixed file.

```
{ "event": "bot.done", "bot_id": "bot_9fJ2p", "recording": { "mp4_url": "https://media.meetstream.ai/..." }, "participants": [ { "name": "Amy Stace", "audio_url": "...", "video_url": "..." }, { "name": "Theo Flynn", "audio_url": "...", "video_url": "..." } ], "transcript_url": "https://api.meetstream.ai/api/v1/transcript/..." }
```

### What you get from every Zoom meeting

- Composed MP4 in grid or speaker view.

- Per-participant audio and video , one stream per person, not a mixed track.

- Real-time media over a secure WebSocket at roughly 200ms, covered on the real-time transcripts page.

- Transcripts through the transcription API , on our in-house engine or on Sarvam, Gemini, Deepgram, AssemblyAI or JigsawStack.

- Participant events for join, leave and admission, plus meeting chat.

- Send Message and Send Image , so the bot can announce itself in the room.

### What Zoom actually enforces

Most integrations that break in production break on a Zoom rule the team did not know about. These are the ones that matter, each linked to Zoom's own documentation.

Cloud recording needs a paid plan. It requires Pro, Business, Education or Enterprise and is not available on Basic. Local recording is available on free accounts , which is a different feature with different limits. If your customers are on free Zoom accounts, native cloud recording is simply not an option and a bot is the only route.

Cloud recording is host-only. It is limited to the host or a co-host , though the host can grant a participant the ability to record from the Participants panel.

The recording disclaimer cannot be turned off for guests. On Basic, Pro and smaller Business accounts it cannot be disabled or customised at all . Larger accounts can disable it for internal participants only; it stays mandatory for external ones on every plan. Participants dialling in by phone always hear an audio prompt. Treat visible notice as a fixed property of Zoom rather than something to design around.

OBF tokens have been mandatory since 2 March 2026. OBF stands for On Behalf Of. It is a short-lived, single-use token that represents your app when a Meeting SDK bot joins as an automated participant, and it carries two constraints worth knowing before you architect around it: a participant must have OAuth-authorized your app, and that authorized user must already be in the meeting for the join to succeed. MeetStream handles this for you, but it explains why a bot cannot always be first into the room.

Waiting Room beats join-before-host. If Waiting Room is on, join before host will not work and the bot waits to be admitted like anyone else.

Consent is contractually yours. Zoom's API License and Terms of Use put the obligation on the developer to ensure all consents have been obtained and all notices given. The recording API page covers the legal position in more detail.

### Zoom's native APIs and where they stop

Zoom has three relevant surfaces, and it is worth knowing what each does before deciding you need a bot.

Cloud Recording API. Fetches recordings after the fact with recording:read . It only returns files that Zoom itself recorded, so it inherits every plan and permission rule above. The webhook download token expires in 24 hours, after which you re-fetch the URL with an OAuth token.

Realtime Media Streams, generally available since June 2025. RTMS delivers per-participant audio, video, transcript, chat and screenshare over a WebSocket with no bot joining at all, and Zoom markets it explicitly as the bot replacement. The trade-offs are real: it needs a user-managed General app on the Zoom Marketplace, it consumes Developer Pack credits, admins can gate it at account, group or user level, and participants see a notice that meeting content is being shared with an app. It is an excellent fit when your users will install your Marketplace app, and no help at all when they will not.

Local archiving token. A 120-second join token for SDK bots that need raw media, requiring the host to be on Pro or higher with archiving enabled.

A bot API sits above all of this. Your code stays the same whether the meeting is on a free account, an enterprise tenant, or a customer who will never install your Marketplace app.

### Getting the bot admitted

The bot appears in the participant list under whatever name you pass, which is deliberate. A named, visible recorder is both better manners and legally safer than a silent one, because several US recording statutes turn on secrecy rather than consent.

If Waiting Room is on, the bot waits for the host. Participant events tell your application whether it was admitted, so you can show a user "waiting to be let in" instead of silently returning nothing. Bots created on demand join in under ten seconds on average, and scheduled bots join at the scheduled time.

### Zoom meeting bot API pricing

A bot costs $0.35 per bot-hour . Transcription adds $0.10 per hour, so a recorded and transcribed meeting is $0.45 per hour all-in , falling toward $0.25 per bot-hour at volume. Thirty days of storage is included rather than metered, and every account starts with $5 of free credit. The pricing page has the full breakdown.

There is no seat licence, no platform fee and no minimum commitment, so the bill tracks usage rather than headcount. If you are comparing against another vendor, the Recall.ai comparison sets out the differences.

### Limits and operational detail

Session length. A bot stays in a Zoom call for four hours by default and can be extended to ten hours through the API, which covers all-hands meetings and full-day sessions.

Concurrency. There is no cap on how many bots run at once and no limit to request an increase to. Capacity planning stays on our side.

Storage and regions. Recordings and transcripts are kept for 30 days, or written into your own S3 bucket so your retention policy governs them from the first byte. Bots run in the United States by default, with EU, Australia, Japan and the Middle East available and other regions on request.

Security. MeetStream is ISO 27001 certified and SOC 2 Type 2 is currently under audit. We do not hold a HIPAA certification and we do not sign Business Associate Agreements today. The security page is kept current.

### How we know this

Zoom is the platform where the rules change most often. The OBF token requirement landed in March 2026 and the SDK key migration follows in June, and both were breaking changes for every bot vendor including us. The constraints on this page are the ones our own bots hit in production, not a summary of someone else's blog post.

MeetStream runs meeting bots on Zoom, Google Meet and Microsoft Teams for more than 30 products in production. Every platform claim on this page links to that platform's own documentation so you can verify it rather than take our word for it, and we have deliberately left out anything we could not source. Where our own position is weaker than a competitor's we say so: we are ISO 27001 certified, SOC 2 Type 2 is under audit rather than complete, and we do not hold a HIPAA certification or sign Business Associate Agreements today.

Platform rules change. If you find something here that no longer matches the vendor documentation, tell us and we will correct it. Written by Sidhdharth Sivasubramanian, co-founder and CEO of MeetStream. Last reviewed 26 August 2026.

### Frequently asked questions

Zoom's own cloud recording is limited to the host or a co-host, though a host can grant a participant the ability to record. A bot records what it receives as a participant, so it does not depend on your user holding host rights, but Zoom still requires participants to be notified.

Zoom cloud recording requires a paid plan. Local recording works on free accounts. A bot-based API does not depend on the host's plan at all, which is the main reason products use one when they cannot control their customers' Zoom accounts.

Yes. The bot appears in the participant list under the name you choose, and Zoom's own recording disclaimer cannot be disabled for external participants on any plan. Phone participants always hear an audio prompt.

OBF stands for On Behalf Of. It is a short-lived, single-use Zoom token required since 2 March 2026 for Meeting SDK bots. MeetStream handles it for you. The constraint worth knowing is that a user who authorized the app must already be in the meeting for the join to succeed.

Sometimes. RTMS gives per-participant media with no bot in the room, which is cleaner, but it needs a user-managed Marketplace app, Developer Pack credits, and admin permission. If your users will install your Zoom app it is a good fit. If they will not, a bot is the only route.

Yes. MeetStream returns a separate audio and video stream per participant, so speaker attribution comes from the platform rather than from a diarization model guessing from a mixed track.

Four hours by default, extendable to ten hours through the API. There is no cap on how many bots run at the same time.

It waits to be admitted, and participant events tell your application that it is waiting. That lets you surface the state to your user rather than returning an empty recording with no explanation.

Bot behaviour depends on the meeting's own admission and participant settings. Tell us your specific configuration and we will confirm before you build against it.

$0.35 per bot-hour for recording, or $0.45 per hour with transcription, dropping toward $0.25 at volume. Thirty days of storage is included and every account starts with $5 of free credit.

### Keep exploring

### Put a bot in your next Zoom call

$5 of free credit, self-serve API keys, and no sales call between you and your first recording.

---

Full site index for agents: https://meetstream.ai/llms.txt
API contract (OpenAPI 3.1): https://meetstream.ai/openapi.json
MCP discovery: https://meetstream.ai/.well-known/mcp