Eleven Realtime Translation
Translate live speech into spoken audio in more than 90 languages through an enterprise WebSocket API.
Eleven Realtime
elevenlabs.io
Build multilingual conversational voice agents with managed speech-to-speech models through ElevenAgents or a WebSocket API.
Visit website
Pricing
Paid
Available on
ElevenAgents web platform, WebSocket API
Category
Ai Audio
What is Eleven Realtime?
Eleven Realtime provides v1 Mini and v1 Max models for conversational voice agents in more than 90 languages. Speech recognition, turn-taking, language processing and voice generation are tuned together behind a single model selection.
Teams can build through ElevenAgents or integrate the separate Realtime API. Mini targets assistants, tutors and interactive characters; Max is designed for complex business interactions and tool-based workflows.
Key features
Two model options
Choose Mini for price and speed, or Max for complex instructions and multi-step tool use.
Managed voice configuration
Realtime mode manages underlying language, speech synthesis and recognition settings while retaining agent prompts, voices, tools, knowledge bases and turn controls.
Compatible WebSocket events
The Realtime API uses an OpenAI-compatible client and server event protocol. Its documented browser flow uses a backend-created, short-lived client secret.
ElevenAgents deployment channels
Use ElevenAgents for phone, web widget, WhatsApp and contact-center deployments. These platform features are separate from the direct Realtime API.
Who is it for?
Conversational tutoring
Education developers could build voice tutors for spoken practice and follow-up questions.
Business voice workflows
Support teams could use Max for conversations that require several tool calls or complex instructions.
Interactive characters
Experience designers could use Mini for conversational characters and lightweight assistants.
How to use Eleven Realtime
- Check that Realtime models are available in your ElevenAgents workspace.
- Create or open an agent, select Realtime voice mode in the Agent tab, and choose Mini or Max.
- Configure the agent's prompt, voice, tools and knowledge base for your task.
- Test the conversation before publishing the configuration for deployment.
- For a direct API integration, follow the WebSocket guide and implement browser audio capture and playback.
Eleven Realtime pricing
Paid
ElevenAgents charges standard call-minute fees plus model token usage; its platform also offers a free entry plan. Direct Realtime API usage is billed for connected audio minutes plus tokens. Workspace rollout restrictions still apply.
Check current pricing ↗What to know before you use it
- Workspace access is rolling out gradually. The API also rejects Realtime configuration for workspaces without access.
- Mini is unavailable with HIPAA or Zero Retention Mode. Max supports those configurations in the US and EU, but is unavailable in India and Singapore residency environments.
- Workflow nodes cannot override managed settings or the token limit. Max requires an output-token limit of at least 1,024 tokens or no limit. Realtime agents cannot transfer to or from Custom-mode agents.
- The direct API uses WebSocket rather than WebRTC. Browser integrations must handle PCM16 audio capture and playback at 24 kHz.
Frequently asked questions
Can I choose the underlying language and speech models?
Realtime mode manages those components. Use Custom mode when you need to choose them individually.
Is this a single fused audio model?
No. The system uses a cascaded architecture with distinct component models; the specific underlying models are not public.
Does the direct API include ElevenAgents features?
No. Choose ElevenAgents when you need its deployment channels, knowledge bases, guardrails, testing or monitoring.