Vonage Video API (formerly OpenTok / TokBox)
Cloud-based WebRTC video API platform enabling embedding of live video, audio, and screen sharing into web and mobile applications. Sessions are created server-side via REST API; tokens are generated to allow clients to connect. Supports multi-party video sessions, archiving (recording to S3 or Vonage cloud), live broadcasting to HLS/RTMP, and signal messaging between session participants. Originally TokBox, acquired by Telefonaktiebolaget LM Ericsson then Vonage (Ericsson acquired Vonage in 2022, brand is now part of Ericsson).
Score Breakdown
⚙ Agent Friendliness
🔒 Security
API Secret must be kept server-side and used to sign client tokens — never expose to browser. Media streams encrypted in transit (DTLS-SRTP for WebRTC). No per-resource scope granularity — API key grants access to all sessions. Archive storage in Vonage cloud is encrypted at rest.
⚡ Reliability
Best When
You need fully customizable WebRTC video embedded into your own application UI, with server-side session control, archiving, and broadcasting — and you are already in the Vonage/Ericsson ecosystem.
Avoid When
You need a pre-built video conferencing experience, you want per-minute pricing without a seat commitment, or Twilio Programmable Video or Daily.co offer better pricing for your usage pattern.
Use Cases
- • Embedding multi-party video consultations in telehealth applications (up to 25 interactive participants)
- • Live interactive broadcasting to thousands of viewers via HLS or WebRTC
- • Recording video sessions to cloud storage for compliance or replay
- • Real-time screen sharing and collaboration features within an application
- • Automated moderation systems that join a session as a server-side subscriber to monitor or record
Not For
- • Standalone consumer video conferencing app (not a SaaS product, purely an API/SDK)
- • Large-scale one-to-many streaming where Mux or AWS MediaLive would be more appropriate
- • Simple audio calls without video where Vonage Voice or Twilio Voice is cheaper
Interface
Authentication
Server API calls use HTTP Basic Auth with API Key and Secret. Client SDK tokens (for end users) are signed JWTs generated server-side using the API Secret — the secret must never be exposed to clients. Token role (subscriber, publisher, moderator) and expiry are encoded in the JWT payload.
Pricing
Pricing is per participant-minute (each client connected to a session counts separately). A 10-minute session with 3 participants = 30 participant-minutes. Archiving costs stack on top of session costs.
Agent Metadata
Known Gotchas
- ⚠ Tokens are short-lived JWTs signed client-side with the API Secret — if an agent generates a token and the client does not connect within the token's TTL (default 24 hours, minimum 30 seconds), the connection will be rejected with a 403 even if the session is valid
- ⚠ Sessions do NOT expire automatically — an agent that creates sessions for each interaction without cleaning them up will accumulate thousands of stale sessions; there is no automatic cleanup and the session list endpoint requires pagination to audit them
- ⚠ Archive (recording) start is asynchronous — the POST /archive call returns 'started' but the actual recording may take several seconds to begin; if the session ends immediately after calling start, the archive may be empty or fail with 'session ended before recording began'
- ⚠ The API Key and Secret are the same credentials used for all Vonage Communications APIs (Video, Voice, SMS); a leaked secret compromises all services, not just video — agents must store and rotate these credentials carefully
- ⚠ WebRTC capability negotiation happens in the browser/native SDK, not the server API; an agent orchestrating a video session cannot programmatically verify that a participant's browser supports the required codecs until the client SDK connection event fires
Alternatives
Full Evaluation Report
Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Vonage Video API (formerly OpenTok / TokBox).
Scores are editorial opinions as of 2026-03-06.