What is Vapi?
Vapi is a developer platform for building real-time voice AI agents. It orchestrates a streaming pipeline of speech-to-text, an LLM, and text-to-speech across telephony providers like Twilio and Vonage.
Resources
Short guides, curated links, a glossary and FAQ for Vapi voice agents and the voice AI ecosystem.
Short guides
Vapi vs Retell vs Bland
All three are developer-first voice agent platforms. Vapi bets on provider flexibility, Retell on natural cadence out of the box, and Bland on enterprise outbound throughput.
How voice-agent pricing works
Vapi's $0.05/min covers orchestration. On top of that you pay STT, LLM, TTS, and telephony pass-through — usually another $0.10–$0.20/min in real workloads.
Reading a status page
A green dot means an endpoint is reachable. It does not guarantee upstream providers (Deepgram, OpenAI, ElevenLabs) are healthy — those failures show up as dropped calls or silent turns.
Why CORS limits browser checks
Most voice AI platforms don't return permissive CORS headers, so browser checks confirm DNS/TCP/TLS handshake only. That's why we report reachability, not full call-path health.
Curated links
Official status pages
Provider pages
Communities
Glossary
- STT
- Speech-to-text; the transcriber that turns caller audio into text (Deepgram, Whisper).
- LLM
- Large Language Model; decides what the agent says next (GPT-4o, Claude, Gemini).
- TTS
- Text-to-speech; turns the LLM's response into audio (ElevenLabs, Aura, PlayHT).
- First-word latency
- Time from caller ending a sentence to hearing the first word of the reply.
- Endpointing
- How the STT decides the caller has finished a turn (silence + energy threshold).
- VAD
- Voice Activity Detection; the low-level signal driving endpointing.
- Barge-in
- Letting the caller interrupt the agent mid-sentence.
- Assistant
- The bundle of prompt, voice, model, and tools that defines an agent in Vapi.
- Tool / function call
- A structured JSON handoff from the LLM to your API mid-call.
- Concurrency
- Number of simultaneous calls your plan can run.
- BYOC
- Bring Your Own Carrier — plug your own telephony provider via SIP.
- BYO provider
- Bring your own OpenAI / ElevenLabs / Deepgram keys instead of managed pricing.
- Warmup
- Pre-loading a TTS voice or LLM route to avoid cold-start on first turn.
- Streaming TTS
- Audio starts as soon as the first token arrives instead of waiting for the full response.
- Turbo / Flash
- Low-latency TTS model variants tuned for real-time conversation.
FAQ
Is Vapi legal in the US?
Yes — Vapi is a developer platform. Voice agent usage rules depend on the jurisdiction and how you use them (recording consent, TCPA, GDPR).
How much does a Vapi call actually cost?
Roughly $0.30–$1.20 for a 3-minute call depending on which STT/LLM/TTS providers you use. See our Vapi pricing per minute guide.
What's the difference between Vapi and Retell?
Vapi gives you fine-grained provider control; Retell ships with tuned natural voices out of the box. Both are strong picks.
Why does this site show reachable but my calls are failing?
Browser checks only see whether public endpoints respond. Failures often come from upstream provider degradation (Deepgram, OpenAI, ElevenLabs) or telephony issues we can't probe from a browser.
Do you place calls or store audio?
No. vapi.health is read-only — we run public reachability checks and link out to each platform.