Speech-to-text, text-to-speech and voice agent APIs.
LiveKit
Open-source realtime platform and Agents framework for voice AI.
SharkScore
#1 in Audio & voice
Why it scores 72.9
LiveKit has a SharkScore of 72.9/100, ranked #1 of 224 in Audio & voice (updated 2026-11-18).
Each part compares LiveKit with other listings in Audio & voice. Evidence covers 87% of the score's weight.
- Adoption
- 59 · 30%
- User reviews
- no data · 25%
- Maintenance
- 100 · 20%
- Trust & transparency
- 80 · 15%
- Momentum
- no data · 10%
Paying never changes this score. How ranking works · method v2026-11-18.2
RankShark review of LiveKit
Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review
We were not paid for this review.
As of 2026-11-18, LiveKit is an open-source framework and cloud platform for voice, video, and physical AI agents, with 1,000 free agent-session minutes monthly.
LiveKit combines an open-source framework for building and testing realtime agents with a cloud platform for deployment, inference, telephony, testing, and observability. The vendor describes it as a developer platform for voice, video, and physical AI agents. [E1]
Best for
- Developers building voice agents who want to deploy them on a managed cloud platform. · home page, pricing page
- Teams building realtime video or robotics applications. · home page
- Teams that want to choose among multiple LLM, speech-to-text, or text-to-speech providers. · home page, pricing page
Not for
- Teams that require maintenance, support, updates, patches, or bug fixes under the Build plan; the terms say LiveKit is not obligated to provide them for that plan. · terms
Key features
- Agent framework. The vendor says the open-source framework supports building and testing agents. · home page
- Cloud deployment. The vendor says LiveKit Cloud can deploy and scale agents on its realtime infrastructure. · home page
- Inference. The vendor says LiveKit Inference provides access to LLM, STT, TTS, and turn-handling models through one API key. · pricing page
- Testing and observability. The vendor lists agent simulations, session recordings, observability events, and session metrics. · pricing page
- Telephony. The vendor lists phone numbers, SIP support, and Twilio and WhatsApp connectors for calls bridged into LiveKit rooms. · home page, pricing page
- Open-source components. The vendor says its media server, SIP service, Egress, Ingress, and SDKs are Apache 2.0 licensed. · security page
Pros
- The Build plan costs $0 per month and includes 1,000 agent-session minutes. · pricing page
- The vendor lists tools for agent simulations, recordings, and turn-by-turn observability. · pricing page
- The vendor lists multiple LLM, STT, and TTS models with per-minute pricing. · pricing page
- The vendor says key components, including its media server and SDKs, are Apache 2.0 licensed open source. · security page
Cons
- The terms say the Build plan does not obligate LiveKit to provide maintenance, support, updates, patches, or bug fixes. · terms
- The terms say published availability or uptime targets are service objectives and do not entitle customers to credits, refunds, or other remedies unless a separate written Enterprise agreement says otherwise. · terms
- Some usage has separate overage charges, so the listed monthly plan price does not cover every usage level or feature. · pricing page
Pricing
| Plan | Price | Limits |
|---|---|---|
| Build | $0/mo | Includes 1,000 agent-session minutes, 5 concurrent agent sessions, 1 agent deployment, and 5,000 WebRTC minutes. |
| Ship | $50/mo | Includes 5,000 agent-session minutes, 20 concurrent agent sessions, 2 agent deployments, and 150,000 WebRTC minutes; overages apply to some usage. |
| Scale | $500/mo | Includes 50,000 agent-session minutes, up to 600 concurrent agent sessions, 4 agent deployments, and 1.5 million WebRTC minutes; overages apply to some usage. |
| Enterprise | Custom | Custom limits; contact sales. |
Checked 18 Nov 2026 · source
- Free plan
- Build costs $0/mo and includes 1,000 agent-session minutes. The homepage also states 1,000 free agent-session minutes monthly. · home page, pricing page
- Free trial
- Not stated (checked 18 Nov 2026)
- Trains on customer data
- LiveKit says it does not use identifiable personal data, customer content, prompts, transcripts, or audio to train or fine-tune its own models. For LiveKit Inference, the vendor says providers are contractually required not to retain or train on inference data. · privacy page
- Data retention
- The vendor says inference data is retained only as long as needed to complete each request. Voice samples may be retained for up to 12 months; samples no longer associated with an active voice clone are deleted after that period. · privacy page
- Data regions
- The vendor says media, recordings, agents, and observability can be pinned to specific regions. The terms describe configuring a pinned region at the customer's request to LiveKit support. · terms, security page
- Data processing agreement
- The security page says a DPA is available. · security page
- AI models used
- The vendor lists LLM, STT, and TTS models from providers including OpenAI, Google, Deepgram, AssemblyAI, Cartesia, and Fish Audio. Listed examples include OpenAI GPT-5.2, Google Gemini 3 Flash, Deepgram Nova-3, and Cartesia Sonic 3. · pricing page
- API
- The vendor lists an inference API and metrics export APIs. · pricing page
- SDKs
- The vendor lists SDKs and provides voice-agent quickstarts for Python or Node.js. Its security page says the SDKs are Apache 2.0 licensed. · home page, security page
- Webhooks
- Not stated (checked 18 Nov 2026)
- Platforms
- The vendor describes users connecting through an app, browser, or phone call, and lists web/mobile starter apps. · home page, pricing page
- Security claims
- Vendor states SOC 2 Type II and says its measures are independently audited against the AICPA's Trust Services Criteria for security, availability, and confidentiality. · security pageThe vendor says WebRTC media uses SRTP with DTLS-SRTP key exchange, SIP supports TLS for signaling and SRTP for media, and data at rest is encrypted with AES-256. · security pageThe vendor says LiveKit Inference and its model providers are bound to zero retention for prompts, audio, and completions. · security pageThe vendor says HIPAA BAAs are available for Scale and Enterprise customers handling protected health information. · security page
What we measured
- GitHub stars · now
- 22,264
- Latest release · now
- v1.13.11 on 2026-10-25
- Package downloads · 30 days
- 33,906,952
- Tranco traffic rank · list of the day
- #153,135
How we tested · sources
- Hourly uptime and response-time checks of the product's URL.
- Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
- We did not sign up, pay, or use the product as a customer.
- home page: https://livekit.com/ · fetched 18 Nov 2026
- pricing page: https://livekit.com/pricing · fetched 18 Nov 2026
- privacy page: https://livekit.com/legal/privacy-policy · fetched 18 Nov 2026
- terms: https://livekit.com/legal/terms-of-service · fetched 18 Nov 2026
- security page: https://livekit.com/security · fetched 18 Nov 2026
Something wrong? Report an error with a public link that shows it; corrections re-run the review.
Top alternatives to LiveKit
See all- 69.3ToolFreemium · from $333.33/mo
- 68.1
Text-to-speech and voice cloning with open models.
ToolFreemium · from $11/mo - 66.9
Make full songs, vocals included, from a prompt.
ToolFreemium · from $8/mo - 66.1
Voice cloning, speech generation and deepfake detection.
ToolPaid · from $280/mo