ElevenLabs

Eleven Realtime v1 Max

eleven-labs/eleven-realtime-v1-max

ElevenLabs' realtime voice model with an OpenAI Realtime-compatible WebSocket API. It reasons before answering for complex instructions and multi-step tool use, supports Zero Retention Mode and HIPAA in the US and EU, and is available only in US and EU data residency environments.

Eleven Realtime v1 Mini

eleven-labs/eleven-realtime-v1-mini

ElevenLabs' realtime voice model with an OpenAI Realtime-compatible WebSocket API.

Eleven v4

eleven-labs/eleven-v4

Eleven v4 is ElevenLabs’ most emotive text-to-speech model for narration, dialogue, and long-form audio in 90+ languages. Write expressive directions into the script: [warm] can set the tone of a greeting, [long pause] can build suspense, and [laughs], [whispers], or [door slams] can add reactions and sound effects at the right moment. The model follows sequences of audio tags, supports multiple speakers, and keeps delivery and voice consistent across longer recordings.

Eleven v4 Turbo

eleven-labs/eleven-v4-turbo

Eleven v4 Turbo brings v4’s expressive speech to real-time conversations and voice agents in 90+ languages. Inline cues such as [warm] for a welcome, [whispers] for a quiet aside, or [laughs] for a natural reaction help shape each turn. Built for streaming agent loops, it can start returning audio while text is still arriving, with about 100 ms median inference latency and about 150 ms median time to first speech. It maintains a consistent voice across long interactions.

Scribe V2

eleven-labs/scribe-v2

Scribe V2 is a speech-to-text model from ElevenLabs for transcription and spoken-language understanding workflows. It is suited to captioning, meeting notes, transcription, and speech analytics pipelines.

Scribe V2 Realtime

eleven-labs/scribe-v2-realtime

Scribe V2 Realtime is a language model from ElevenLabs for chat, writing, analysis, and general-purpose automation.