Which voice AI agent platforms have Latency claim?
A latency figure the vendor publishes on an official page. These are vendor claims, not our measurements.
4 Yes1 Partial6 Unknown
Verified Yes: Vapi, Retell AI, LiveKit Agents, Vaani.
| Platform | Status | What the vendor says | Source |
|---|---|---|---|
| LiveKit AgentsLiveKit | Yes | 100ms network "100ms global latency" refers to LiveKit's network, not full agent response. | Source |
| Retell AIRetell AI | Yes | ~600ms Home page claims "~600ms latency". | Source |
| VaaniVaani | Yes | p90 ~450ms "p90 latency of ~450ms". | Source |
| VapiVapi | Yes | <500ms avg Home page claims "<500ms average latency". | Source |
| SynthflowSynthflow | Partial | Sub-100ms telephony "Sub-100 ms latency" is claimed for its telephony layer, not end-to-end response time. | Source |
| Bland AIBland | Unknown | No statement found on the official pages we checked. | — |
| Cartesia AgentsCartesia | Unknown | No statement found on the official pages we checked. | — |
| Deepgram Voice Agent APIDeepgram | Unknown | No statement found on the official pages we checked. | — |
| ElevenLabs AgentsElevenLabs | Unknown | Vendor says "low latency" but publishes no figure on the pages we checked. | — |
| OpenAI Realtime APIOpenAI | Unknown | No statement found on the official pages we checked. | — |
| UltravoxUltravox | Unknown | Home page discusses latency but gives no figure. | — |
Checked Nov 4, 2026. Filter the matrix by this feature →
Other features
Free, when things change
Get the voice-agent changelog
Price cuts, new MCP and compliance support, and new platforms, in one short email when the matrix changes.