Real-time voice AI — the kind of app where you talk and an AI answers back almost instantly — sounds simple until you try to build it. You need speech recognition, a language model, speech generation, and all of it has to stream fast enough that the conversation doesn't feel like a walkie-talkie. Micdrop and Vapi both remove that plumbing, but for different jobs: Micdrop is a free, open-source toolkit you wire into your own web app and host wherever you want; Vapi is a managed platform built specifically to turn that same pipeline into an AI that can actually answer a phone call, billed per line.
The short version
| Micdrop | Vapi | |
|---|---|---|
| Format | Open-source SDK (npm/GitHub), you self-host | Managed SaaS platform |
| Built for | Real-time voice conversation inside your own app or browser | Phone-answering AI agents (real telephony/PSTN) |
| Pricing | Free, MIT-licensed SDK — you pay your chosen AI providers directly | Freemium: ~$0.05/min hosting + $10 per concurrent line/month |
| AI provider choice | Bring your own (OpenAI, ElevenLabs, Mistral, Cartesia, Gladia, Gradium...) | Built-in pipeline; can bring your own API key to control model cost |
| Hosting / data residency | Self-hosted, can be kept fully within the EU for GDPR | Vapi's managed cloud |
| Real phone numbers | Not built-in — browser/app-first | Yes, this is the core feature |
| Best fit | Developers who want full control and no vendor lock-in | Teams who want a working phone voice agent fast, without building the pipeline |
Micdrop — the open-source toolkit you wire up yourself
Micdrop is an open-source developer toolkit for building real-time voice conversation — the kind you'd have with a phone-support AI — directly in a web browser, without gluing together five separate services by hand. It's agnostic about which AI providers you use, so nothing locks you into one vendor, and it supports a fully EU-based, GDPR-compliant deployment for teams that need to keep voice data on European infrastructure. It's also built for low-latency streaming with proper voice activity detection, so it doesn't feel like a walkie-talkie.
Who it's for: developers building AI voice conversation features into their own web app, and teams with real GDPR requirements around where voice data lives.
Pricing: the SDK itself is free and MIT-licensed; the real cost is whichever third-party AI providers you connect (OpenAI, ElevenLabs, Mistral, Cartesia, Gladia or Gradium).
Honest limits: you're responsible for managing your own third-party API keys, it requires real TypeScript/Node.js expertise to integrate properly, and there's no bundled pricing — your total cost depends entirely on which providers you pick.
Vapi — the managed platform for phone-answering agents
Vapi is a developer platform that wires together speech recognition, an AI model and speech generation into a working phone-answering voice agent — the part Micdrop leaves to you, Vapi has already built and hosts for you, including the actual telephony (PSTN) connection to real phone numbers. It targets sub-500ms latency for natural-feeling calls and lets you bring your own model API key if you want more control over cost and quality.
Who it's for: SaaS developers and enterprise teams that need a phone-based AI agent (support lines, outbound calling) working quickly, without building the speech pipeline themselves.
Pricing: freemium, with a Build tier around $0.05/min hosting plus $10 per concurrent line per month, and a custom-priced Scale tier for larger contracts.
Honest limits: Vapi competes closely with Retell AI and Bland AI with limited fundamental differentiation between them, the per-concurrent-line pricing adds real cost as call volume grows, and underlying model provider costs are a separate variable you still have to budget for.
Pick Micdrop if…
…you're building a voice feature inside your own app or browser, you want zero vendor lock-in and full control over which AI providers you use, and you're comfortable doing the TypeScript/Node.js integration yourself.
Pick Vapi if…
…you specifically need an AI that can answer or make real phone calls, want it working without building the STT/LLM/TTS pipeline yourself, and per-line pricing is acceptable for your call volume.
In practice the overlap is small: Micdrop doesn't do telephony at all, and Vapi isn't meant to be embedded as a lightweight browser SDK. The real decision is usually made for you by the use case — phone number reachability points to Vapi, in-app voice UX with full control points to Micdrop.