Grok Voice API
Realtime speech-to-speech API on Grok Voice 3.0 for building voice agents, at $0.08 per minute, with SIP calls, tool use and OpenAI Realtime compatibility.
- Pricing at a glance
- $0.08/min speech-to-speech; TTS $15/M characters; STT from $0.10/hour
- Starting price
- From $0.08/min
- Setup
- Developer build
- Pricing model
- Per Minute
- Try it
- Public web demo
Voice agents in the Grok API Console; a Console account is needed, and API usage is billed.
At a glance
Calls & telephony
- DetailCalls
Inbound only
- SupportedHuman handoff
Yes
- Not publishedPhone numbers included
Not published
- SupportedBring your own carrier
Yes
- SupportedSIP trunking
Yes
Bring your own SIP trunk number; docs mention SpaceXSI numbers but say they can't be provisioned via API
Trust & access
- With conditionsHIPAA
Enterprise plans
BAA after a questionnaire and SpaceXSI review
- SupportedSOC 2
Yes
API page lists SOC 2; report type not stated
- SupportedGDPR
Yes
DPA incorporated into the Enterprise Terms
- DetailLanguages
20+ languages (product page says 25+)
- Not supportedFree to start
Paid only
API usage is billed; the Console Playground can be tried before adding billing
- SupportedDeveloper API
Yes
Morak scorecard
How we score →- Ease of setupDeveloper build
Expect developers to build call logic, integrations and deployment.
- Pricing transparencyRates published
Per-minute rates are public, so you can estimate cost at your volume.
- Try before you buyWeb demo
A public web demo shows the experience, not your own setup.
- Call & telephony controlModerate
Documented: inbound, human handoff, bring-your-own carrier and SIP.
- Compliance readinessGood
Published: SOC 2, GDPR and HIPAA on enterprise plans.
- Integration coverage2 of 4 covered
Connects to phone system and automation or API. Not listed: customer records (CRM, EHR or POS) and calendar.
Typical score among developer APIs
Build on Grok's speech-to-speech model at one per-minute price
What you get
Stream calls to a reasoning speech-to-speech model with built-in web and X search, MCP and your own functions, over SIP, WebSocket or LiveKit, for $0.08 a minute.
What to account for
Outbound calling isn't documented, you can't swap the speech or LLM parts, and there's no published consent rule for cloned voices.
Best for
- Developer teams building realtime voice agents
- Teams moving off the OpenAI Realtime API
- Agents that need live web or X search
- One flat per-minute speech-to-speech price
Not a fit for
- Non-technical teams wanting a finished receptionist
- Outbound calling, which the SIP docs don't cover
- Teams that want to choose speech and LLM providers separately
What the platform provides
- Speech-to-speech agents
- Grok Voice 3.0 over a WebSocket, with reasoning, turn detection, web and X search, MCP and your own functions.
- Phone and app connections
- SIP with your own trunk, WebSocket or LiveKit, with G.711 audio, DTMF and call transfer.
Connections & compatibility
Telephony
- SIP
- Twilio
AI models & voice
- LiveKit
- OpenAI Realtime clients
Automation & APIs
- WebSocket
- MCP
Other
- Web and X search
Data access & permissions
What an agent built on Grok Voice API can touch once deployed. These are the first questions a security review will ask.
- Reads customer dataYes
- Writes to your systemsYes
- Sends messagesYes
- Triggers workflowsYes
- Records callsNo
- Needs API keysYes
- Human approval stepYes
SpaceXSI's enterprise terms delete API inputs and outputs within 30 days unless agreed otherwise or held for safety or legal reasons, and a zero data retention option keeps no content after the response. A DPA covers GDPR, BAAs are offered after a questionnaire and review, and the API page lists SOC 2 without a report type. Voice cloning has no published rule on whose voice may be cloned or how consent is checked, though the acceptable use policy bans deceptively impersonating a real person: clone only voices you have permission to use. Custom voices are only offered in the US, excluding Illinois. The SIP docs say nothing on calling consent, so you own calling-law compliance. Morak has not audited these claims.
Cost at your volume
Speech to Speech with grok-voice-3.0 costs $0.08 per minute ($4.80 an hour) plus $0.004 per text input, the same as the deprecated Grok Voice 2.0. Text-to-speech is $15 per million characters, and speech-to-text $0.10 an hour (REST) or $0.20 an hour (streaming). The general tool price list has web search at $5 per 1,000 calls and collections search at $2.50 per 1,000, but the pricing page doesn't say how tools are billed inside voice sessions. Up to 30 custom voices are free in the console; creating them through the API needs an Enterprise plan. Carrier costs are separate, and API usage is prepaid or invoiced.
Before committing
Interrupt the agent on a noisy phone line with reasoning on and off, fail a tool call, then transfer the caller.
- How do latency and accuracy hold up on 8 kHz phone audio in our languages, with reasoning on and off?
- How are web search and other tools billed inside a voice session?
- Can agents place outbound calls, and can you provide phone numbers?
- What consent do you require before a custom voice is cloned?
How it compares
Side by side with two alternatives buyers weigh against it, on the same facts and scorecard.
| Detail | Grok Voice API This productRealtime speech-to-speech API on Grok Voice 3.0 for building voice agents, at $0.08 per minute, with SIP calls, tool use and OpenAI Realtime compatibility. | A different approach Realtime speech-to-speech API for building low-latency voice agents over WebRTC, WebSocket, or SIP. Why compare: Compare OpenAI's Realtime API, which Grok's API mirrors, if you want OpenAI's models behind the same kind of client. | Similar product Conversational voice platform focused on expressive speech, realtime interaction, and prosody-aware responses. |
|---|---|---|---|
| Starting price | From $0.08/min | From $32/M tokens | From $0.04/min |
| Free to start | Paid only | Not published | Free plan |
| Calls | Inbound | Inbound & outbound | Not a phone product |
| HIPAA | Enterprise plans | Enterprise plans | Enterprise plans |
| Morak scorecard How we score | |||
| Ease of setup | 2 of 5, Developer build | 2 of 5, Developer build | 2 of 5, Developer build |
| Pricing transparency | 5 of 5, Rates published | 5 of 5, Plans published | 5 of 5, Rates published |
| Try before you buy | 3 of 5, Web demo | 3 of 5, Web demo | 5 of 5, Free plan |
| Call & telephony control | 3 of 5, Moderate | 4 of 5, Broad | — |
| Compliance readiness | 4 of 5, Good | 5 of 5, Strong | 5 of 5, Strong |
| Integration coverage | 3 of 5, 2 of 4 covered | 3 of 5, 2 of 4 covered | 2 of 5, 1 of 3 covered |