Kyutai Moshi vs VoxAI

A side-by-side look at pricing, capabilities, pros, cons, and our editorial scores.

	Kyutai Moshi Audio	VoxAI Audio
Tagline	Open-source, full-duplex speech-to-speech foundation model with sub-200ms latency	Local macOS conversation recorder with live AI copilot and speaker-labelled transcripts
Category	Audio	Audio
Pricing	Free· Free and open source. Models under CC-BY 4.0, code under MIT (Python) / Apache 2.0 (Rust). Self-hosted only — you pay your own compute (24GB+ GPU for PyTorch, or Apple Silicon via MLX).	Paid· FREE TRIAL: US$0 · EARLY BIRD · REGULAR: US$19.90
Model	Moshi (7B-class speech-text foundation model) + Mimi neural audio codec, in-house by Kyutai	Bring-your-own via MCP (Claude, GPT, or other MCP-compatible clients)
Editorial score	—	—
Use cases	Real-time voice assistant prototypesResearch on full-duplex spoken dialogueOn-device voice interaction on Apple Silicon via MLXLow-latency conversational agents behind WebSocketNeural audio codec experimentation with MimiSelf-hosted voice interface for privacy-sensitive appsSpeech tokenizer for downstream audio LLM trainingInterruptible in-car or wearable voice UX	Real-time meeting transcription with speaker labelsLive AI copilot during sales calls and negotiationsInterview capture and structured note generation1:1 and standup summarisationQualitative research transcriptionPersonal searchable voice memory archiveFeeding conversation history to Claude Code or Cursor via MCPPost-call action item and follow-up extractionMarkdown export into Notion or Obsidian knowledge bases
Pros	Truly full-duplex — handles interruptions, overlap and back-channels rather than rigid turn-taking Sub-200ms practical latency on a single L4 GPU, well below third-party voice APIs Fully open weights (CC-BY 4.0) plus MIT/Apache code — self-host with no per-minute billing Ships with Mimi, a streaming neural audio codec that beats SpeechTokenizer and SemantiCodec Multiple inference backends: PyTorch for research, Rust/Candle for production, MLX for on-device Mac/iPhone Inner-monologue text prediction gives you a transcript alongside the audio stream for free	One-time ~$20 purchase with permanent license for two Macs — no recurring subscription Local-first: recordings and transcripts stay on the Mac in a permanent archive Live AI copilot reads the growing transcript and offers real-time suggestions during the call Automatic speaker diarisation with per-speaker labels Four purpose-built modes for meetings, phone calls, in-person chats, and voice sessions Exposes conversations via a local MCP server so Claude Code, Codex, or Cursor can query them Clean Markdown export for pasting into Notion, Obsidian, or a CRM 10 full-feature sessions free with no signup required
Cons	English-only voices at launch — no multilingual support out of the box Knowledge and reasoning quality trail top text LLMs; it's a 7B-class model, not GPT-4o Voice Requires a 24GB+ GPU for the reference PyTorch build; on-device is only viable via MLX on Apple Silicon No hosted API or SaaS tier — you own the ops, scaling and safety filtering Only two fixed synthetic voices (Moshiko/Moshika); no voice cloning or speaker conditioning in the release	macOS only — no Windows, Linux, iOS, web, or mobile client Requires wiring up an external AI provider for the copilot and post-call generation Solo-app model: no team workspace, shared library, or admin console Small indie product from a single developer domain — long-term support is not guaranteed MCP integration assumes users are already comfortable with Claude Code / Cursor tooling No native CRM, calendar, or Zoom bot integrations — capture is whatever the Mac hears
Website	kyutai.org	voxai.ethanflow.com

Pick Kyutai Moshi if

✅ Truly full-duplex — handles interruptions, overlap and back-channels rather than rigid turn-taking
✅ Sub-200ms practical latency on a single L4 GPU, well below third-party voice APIs
✅ Fully open weights (CC-BY 4.0) plus MIT/Apache code — self-host with no per-minute billing
✅ Ships with Mimi, a streaming neural audio codec that beats SpeechTokenizer and SemantiCodec

Pick VoxAI if

✅ One-time ~$20 purchase with permanent license for two Macs — no recurring subscription
✅ Local-first: recordings and transcripts stay on the Mac in a permanent archive
✅ Live AI copilot reads the growing transcript and offers real-time suggestions during the call
✅ Automatic speaker diarisation with per-speaker labels

Compare a different pair →