インフラ
LiveKit
オープンソースのWebRTCリアルタイム音声・映像基盤 — ChatGPT音声モードのエンジン。
livekit.io · 掲載≠推奨 · 2026-09-03 再確認
- 創業者
- Russ d’Sa & David Zhao
- 創業
- 2021
- 評価額
- $1B · Series C led by Index (Jan 2026)
- 顧客
- OpenAI, xAI, Salesforce, Tesla, …
- この波での役割
- Reported WebRTC layer for fal.live rooms
LiveKit とは?
音声・映像エージェント向けのオープンソースWebRTCリアルタイムA/V基盤。最も有名なのは OpenAI の ChatGPT 音声モードの裏側として使われていることで、2026年1月に10億ドル評価に。この波の中では fal.live 型ルームの「報道ベース」のトランスポート層。
LiveKit のできること
- Open-source WebRTC realtime audio/video infrastructure
- LiveKit Agents framework for voice and video multimodal agents
- Managed LiveKit Cloud alongside the self-hostable open-source core
- Used by voice products, 911 operators, telehealth and live rooms
LiveKit は誰向け?
- Building voice-mode products on top of a proven realtime transport
- Routed live A/V for agents and human-plus-agent apps
- Real-time generative video rooms where a stream needs WebRTC plumbing
LiveKit の長所と短所
長所
- Proven at very large scale — reported as the engine behind OpenAI’s ChatGPT voice mode
- Open source with a managed cloud, so you can start free and scale
- Unicorn funding ($1B, Jan 2026) backs long-term investment
短所
- An infrastructure layer, not an end product — you still build on it
- Self-hosting at scale is real engineering work
- Any specific link to generative-video rooms (e.g. fal.live) is reported, not documented
世間の声
公開報道からの引用 — 収集日 2026-09-03。
“LiveKit powers OpenAI’s ChatGPT voice mode.”
“…the raise of $100 million in funding at a $1 billion valuation.”
よくある質問
What is LiveKit?
Open-source WebRTC realtime audio/video infrastructure plus an agents framework for building voice and video AI applications.
Is LiveKit open source?
Yes — the core is open source and can be self-hosted; LiveKit Cloud is the managed option.
Who uses LiveKit?
TechCrunch reports OpenAI (ChatGPT voice mode), xAI, Salesforce, Tesla, plus 911 operators and mental-health providers.
Is LiveKit related to fal.live?
Any fal.live↔LiveKit link is reported rather than documented as of Sep 2026.
インフラの他の選択肢
3
fal.ai
低遅延の生成メディア推論プラットフォーム —「モデルから直接ライブ配信」の波の基盤。fal.ai
Daydream
Open-source, local-first runtime for realtime generative-video pipelines — run a video-diffusion node graph and stream it over WebRTC.livepeer.org