Vapi
Developer platform for building, testing, and deploying AI voice agents that can handle real phone calls at scale
Pay-as-you-go from $0.05/min; no monthly minimum
Overview
Vapi is the infrastructure layer for AI voice agents — handling all the complexity of real-time telephony, speech recognition, LLM integration, and text-to-speech so developers can focus on building the agent logic. It powers everything from customer service bots to appointment schedulers to outbound sales dialers.
Key Features
- End-to-end voice agent infrastructure: STT, LLM, TTS in under 500ms latency
- Inbound and outbound phone call handling
- Real phone number provisioning (US and international)
- Function calling: let voice agents take actions during a call
- Multiple STT and TTS provider support (Deepgram, ElevenLabs, etc.)
- Call recording, transcription, and analytics dashboard
- SDK for JavaScript, Python, iOS, and Android
Pricing: Pay-as-you-go at ~$0.05/min; no monthly minimum.
Pros
- Best infrastructure for production voice agents at scale
- Sub-500ms latency makes conversations feel natural
- Supports multiple STT/TTS providers for customization
- No monthly minimum — pay only for usage
Cons
- Developer-focused — requires coding to build agents
- Production costs can scale significantly with volume
- Complex pipeline of STT + LLM + TTS requires tuning for each use case
Tags
Product Updates
Similar Tools
AssemblyAI
Speech AI API for transcription, speaker detection, sentiment analysis, and audio intelligence — used by developers to build audio-powered applications
Retell AI
Platform for building and deploying human-like AI voice agents for customer service, sales, and appointment scheduling at scale
ChatGPT
OpenAI's model family featuring GPT-4o, o1, o3, and DALL-E
Claude
Anthropic's frontier model family featuring Opus, Sonnet, and Haiku
