Skip to main content

LiveKit Inference

Access the best AI models for voice agents, included in LiveKit Cloud.

On this page

Overview

LiveKit Inference provides access to many of the best models and providers for voice agents, including STT, LLM, TTS, and realtime models from OpenAI, Google, AssemblyAI, Deepgram, Cartesia, Fish Audio, Inworld, and more. LiveKit Inference is included with LiveKit Cloud and doesn't require additional plugins. It's zero data retention by default. Your prompts, audio, and model outputs are never stored or used to train models. To learn more about supported models and configuration options, see the LLM, STT, TTS, and realtime topics.

By default, LiveKit Inference routes requests to model providers without regional restrictions. To restrict requests to models served from your project's data region, such as for EU data residency, enable region restriction.

To learn more about LiveKit Inference, see the blog post Introducing LiveKit Inference: A unified model interface for voice AI .

For LiveKit Inference models, use the inference module classes in your AgentSession:

from livekit.agents import AgentSession, inference
session = AgentSession(
stt=inference.STT(
model="deepgram/flux-general",
language="en"
),
llm=inference.LLM(
model="google/gemma-4-31b-it",
),
tts=inference.TTS(
model="fishaudio/s2.1-pro",
voice="fa4c9eb3dccc4806b382b40d61c6b10a",
),
)
import { AgentSession, inference } from '@livekit/agents';
const session = new AgentSession({
stt: new inference.STT({
model: "deepgram/flux-general",
language: "en"
}),
llm: new inference.LLM({
model: "google/gemma-4-31b-it",
}),
tts: new inference.TTS({
model: "fishaudio/s2.1-pro",
voice: "fa4c9eb3dccc4806b382b40d61c6b10a",
}),
});

String descriptors

As a shortcut, you can pass a model descriptor string directly instead of using the inference classes. This is a convenient way to get started quickly.

from livekit.agents import AgentSession
session = AgentSession(
stt="assemblyai/assemblyai-stt-3-7-realtime-agent",
llm="google/gemma-4-31b-it",
tts="fishaudio/s2.1-pro:fa4c9eb3dccc4806b382b40d61c6b10a",
)
import { AgentSession } from '@livekit/agents';
const session = new AgentSession({
stt: "assemblyai/assemblyai-stt-3-7-realtime-agent",
llm: "google/gemma-4-31b-it",
tts: "fishaudio/s2.1-pro:fa4c9eb3dccc4806b382b40d61c6b10a",
});

For detailed parameter references and model-specific options, see the individual model guides for LLM, STT, and TTS.

Provider pinning deprecation

Support ends on March 1, 2027

Provider pinning is deprecated. Support for the provider option in LiveKit Inference LLM, Responses, realtime, and duplex APIs ends on March 1, 2027.

Pinning restricts requests to one host and prevents fallback to other hosts serving the same model. Remove the provider argument in Python or the provider option in Node.js to let LiveKit route requests and use fallback providers. If you call the inference endpoint directly, remove the X-LiveKit-Inference-Provider header.

Keep the full model ID, including its provider prefix. For example, change inference.LLM(model="openai/gpt-oss-120b", provider="groq") to inference.LLM(model="openai/gpt-oss-120b"). In Node.js, change new inference.LLM({ model: "openai/gpt-oss-120b", provider: "groq" }) to new inference.LLM({ model: "openai/gpt-oss-120b" }).

If your application requires a specific host, use that provider's plugin with your own credentials. For geographic requirements, configure region restriction.

Authentication

LiveKit Inference uses your LiveKit Cloud credentials. You don't need a separate API key for each model provider. Set LIVEKIT_API_KEY and LIVEKIT_API_SECRET in your .env file, or pass api_key and api_secret directly to an inference class. In Node.js, these parameters are apiKey and apiSecret.

Zero data retention

LiveKit Inference is zero data retention (ZDR) by default. Your prompts, audio, and model outputs pass through only to generate a response. Neither LiveKit nor the underlying model providers log, store, or train on your data. This applies to every LLM, STT, TTS, and realtime model, on every plan, with no configuration required.

What zero data retention covers

Zero data retention applies to the data you send to model providers through LiveKit Inference. It's separate from agent observability, which you can enable to retain session data in LiveKit Cloud, and from custom voices, which store the voice sample you provide so you can reuse the clone across providers.

Region restriction

By default, LiveKit Inference can route requests to a model provider in any region. To keep inference requests within your project's data region, enable region restriction.

For new projects with agents hosted in the EU, region restriction is enabled by default. To enable it for an existing project, sign in to LiveKit Cloud  and select Project → Settings. In the Data and privacy section, set Inference region restriction to Enabled.

When enabled, LiveKit Inference only routes requests to models with an endpoint in your project's data region. For a project with the EU data region, this limits inference to models served from the EU. Models without an endpoint in your data region are shown as unavailable in the Agent Builder.

Applies only to agents deployed to LiveKit Cloud

Region restriction applies only to requests from agents deployed to LiveKit Cloud. Requests from an agent that runs anywhere else, such as on your own machine during development or on your own servers, aren't restricted, even when the setting is enabled.

Region restriction applies only to models routed through LiveKit Inference. Provider plugins send requests directly to the provider, so you must configure the provider's regional endpoint separately. To configure EU data residency across the rest of your stack, see EU data residency.

Data residency

Some models offer a dedicated endpoint in the EU, in addition to the global endpoint. A checkmark in the EU endpoint column of the model tables means the model has an EU endpoint, which lets you keep inference requests in the EU with region restriction.

Models

The following tables list all models currently available through LiveKit Inference.

Pricing

See the latest pricing for all LiveKit Inference models.

Large language models (LLM)

Browse the available large language models by model family.

Recommended for voice agents

Gemma 4 31B is the recommended default LLM. It's a latency-optimized, open-weight model served on LiveKit's infrastructure.

A checkmark in the EU endpoint column means the model has a dedicated EU endpoint, in addition to the global endpoint. To learn more, see Data residency.

Hosted by LiveKit

Model nameModel IDProvidersEU endpoint
Gemma 4 31B
google/gemma-4-31b-it
  • livekit
EU endpoint

DeepSeek

Model nameModel IDProvidersEU endpoint
DeepSeek-V4 Pro
deepseek-ai/deepseek-v4-pro
  • vercel
No EU endpoint
DeepSeek-V4.1 Flash
deepseek-ai/deepseek-v4.1-flash
  • baseten
No EU endpoint
DeepSeek-V3
deepseek-ai/deepseek-v3
  • baseten
    Retired
No EU endpoint
DeepSeek-V3.1
deepseek-ai/deepseek-v3.1
  • baseten
    Retired
No EU endpoint
DeepSeek-V3.2
deepseek-ai/deepseek-v3.2
  • deepseek
    Retired
No EU endpoint

Gemini

Model nameModel IDProvidersEU endpoint
Gemini 3 Flash
google/gemini-3-flash-preview
  • google
No EU endpoint
Gemini 3.1 Flash Lite
google/gemini-3.1-flash-lite
  • google
EU endpoint
Gemini 3.1 Pro
google/gemini-3.1-pro-preview
  • google
No EU endpoint
Gemini 3.5 Flash
google/gemini-3.5-flash
  • google
EU endpoint
Gemini 3.5 Flash Lite
google/gemini-3.5-flash-lite
  • google
EU endpoint
Gemini 3.8 Flash
google/gemini-3.8-flash
  • google
EU endpoint
Gemini 4 Argon
google/gemini-4-argon
  • google
EU endpoint
Gemini 4 Hydrogen
google/gemini-4-hydrogen
  • google
EU endpoint
Gemini 4 Neon
google/gemini-4-neon
  • google
No EU endpoint
Gemini Decision 1
google/gemini-decision-1
  • google
No EU endpoint
Gemini 3.6 Flash
google/gemini-3.6-flash
  • google
    Deprecated
    Retires
EU endpoint
Gemini 3.7 Flash
google/gemini-3.7-flash
  • google
    Deprecated
    Retires
EU endpoint
Gemini 2.0 Flash
google/gemini-2.0-flash
  • google
    Retired
No EU endpoint
Gemini 2.0 Flash-Lite
google/gemini-2.0-flash-lite
  • google
    Retired
No EU endpoint
Gemini 2.5 Flash
google/gemini-2.5-flash
  • google
    Retired
EU endpoint
Gemini 2.5 Flash-Lite
google/gemini-2.5-flash-lite
  • google
    Retired
EU endpoint
Gemini 2.5 Pro
google/gemini-2.5-pro
  • google
    Retired
EU endpoint
Gemini 3 Pro
google/gemini-3-pro-preview
  • google
    Retired
No EU endpoint

meta

Model nameModel IDProvidersEU endpoint
Muse Spark 1.3
meta/muse-spark-1.3
  • meta
No EU endpoint
Muse Wonder 1.0
meta/muse-wonder-1.0
  • meta
No EU endpoint

Kimi

Model nameModel IDProvidersEU endpoint
Kimi K2.6
moonshotai/kimi-k2.6
  • vercel
No EU endpoint
Kimi K2 Instruct
moonshotai/kimi-k2-instruct
  • baseten
    Retired
No EU endpoint
Kimi K2.5
moonshotai/kimi-k2.5
  • baseten
    Retired
No EU endpoint

OpenAI

Model nameModel IDProvidersEU endpoint
ChatGPT Latest
openai/chat-latest
  • azure
  • openai
No EU endpoint
GPT-5
openai/gpt-5
  • azure
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5 mini
openai/gpt-5-mini
  • azure
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5 nano
openai/gpt-5-nano
  • azure
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5.2
openai/gpt-5.2
  • azure
  • openai
EU endpoint
GPT-5.4
openai/gpt-5.4
  • azure
  • openai
EU endpoint
GPT-5.4 mini
openai/gpt-5.4-mini
  • azure
  • openai
EU endpoint
GPT-5.5
openai/gpt-5.5
  • azure
  • openai
EU endpoint
GPT-5.6 Luna
openai/gpt-5.6-luna
  • azure
  • openai
EU endpoint
GPT-5.6 Sol
openai/gpt-5.6-sol
  • azure
  • openai
EU endpoint
GPT-5.6 Terra
openai/gpt-5.6-terra
  • azure
  • openai
EU endpoint
GPT-6 Luna
openai/gpt-6-luna
  • azure
  • openai
EU endpoint
GPT-6 Luna (Decisions)
openai/gpt-6-luna-decisions
  • openai
EU endpoint
GPT-6 Sol
openai/gpt-6-sol
  • azure
  • openai
EU endpoint
GPT-6.1 Sol
openai/gpt-6.1-sol
  • azure
  • openai
EU endpoint
GPT-6.1 Terra
openai/gpt-6.1-terra
  • openai
EU endpoint
GPT OSS 120B
openai/gpt-oss-120b
  • baseten
  • groq
No EU endpoint
GPT-4.1
openai/gpt-4.1
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-4.1 mini
openai/gpt-4.1-mini
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-4.1 nano
openai/gpt-4.1-nano
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-4o
openai/gpt-4o
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-4o mini
openai/gpt-4o-mini
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5.1
openai/gpt-5.1
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5.4 nano
openai/gpt-5.4-nano
  • azure
    Deprecated
    Retires
  • openai
    Deprecated
    Retires
EU endpoint
GPT-5.1 Chat
openai/gpt-5.1-chat-latest
  • azure
    Retired
  • openai
    Retired
No EU endpoint
GPT-5.2 Chat
openai/gpt-5.2-chat-latest
  • azure
    Retired
  • openai
    Retired
No EU endpoint
GPT-5.3 Chat
openai/gpt-5.3-chat-latest
  • azure
    Retired
  • openai
    Retired
No EU endpoint

SpaceXAI

Model nameModel IDProvidersEU endpoint
Grok 4.20
xai/grok-4.20-0309-non-reasoning
  • xai
No EU endpoint
Grok 4.20 Reasoning
xai/grok-4.20-0309-reasoning
  • xai
No EU endpoint
Grok 4.20 Multi-Agent
xai/grok-4.20-multi-agent-0309
  • xai
No EU endpoint
Grok 4.3
xai/grok-4.3
  • xai
No EU endpoint
Grok 4.5
xai/grok-4.5
  • xai
No EU endpoint
Grok 4.6
xai/grok-4.6
  • xai
No EU endpoint
Grok 4.7
xai/grok-4.7
  • xai
No EU endpoint
Grok 4.1 Fast
xai/grok-4-1-fast-non-reasoning
  • xai
    Retired
No EU endpoint
Grok 4.1 Fast Reasoning
xai/grok-4-1-fast-reasoning
  • xai
    Retired
No EU endpoint
Retired models
A Retired badge means that provider no longer serves the model. A model that is retired by every listed provider is no longer accessible. Switch to a currently available model.

Speech-to-text (STT)

Browse the available speech-to-text models by model family.

A checkmark in the EU endpoint column means the model has a dedicated EU endpoint, in addition to the global endpoint. To learn more, see Data residency.

AssemblyAI

Model nameModel IDLanguagesEU endpoint
STT 3.7 Realtime Agent
assemblyai/assemblyai-stt-3-7-realtime-agent
enen-USen-GBen-AUen-CAen-INen-NZeses-ESes-MXes-ARes-COes-CLes-PEes-VEes-ECes-GTes-CUes-BOes-DOes-HNes-PYes-SVes-NIes-CRes-PAes-UYes-PRfrfr-FRfr-CAfr-BEfr-CHdede-DEde-ATde-CHitit-ITit-CHptpt-BRpt-PTtrnlsvnodafihiviarhejaurzhrukocaglroetfayueafmrzuxhnn
EU endpoint
Universal-3.6 Pro
assemblyai/universal-3-6-pro
enen-USen-GBen-AUen-CAen-INen-NZeses-ESes-MXes-ARes-COes-CLes-PEes-VEes-ECes-GTes-CUes-BOes-DOes-HNes-PYes-SVes-NIes-CRes-PAes-UYes-PRfrfr-FRfr-CAfr-BEfr-CHdede-DEde-ATde-CHitit-ITit-CHptpt-BRpt-PTtrnlsvnodafihiviarhejaurzhrukocaglroetfayueafmrzuxhnn
EU endpoint
Universal-3.5 Pro
assemblyai/universal-3-5-pro
enen-USen-GBen-AUen-CAen-INen-NZeses-ESes-MXes-ARes-COes-CLes-PEes-VEes-ECes-GTes-CUes-BOes-DOes-HNes-PYes-SVes-NIes-CRes-PAes-UYes-PRfrfr-FRfr-CAfr-BEfr-CHdede-DEde-ATde-CHitit-ITit-CHptpt-BRpt-PTtrnlsvnodafihiviarhejaurzh
EU endpoint
Universal-Streaming
assemblyai/universal-streaming
enen-US
EU endpoint
Universal-Streaming-Multilingual
assemblyai/universal-streaming-multilingual
multienen-USen-GBen-AUen-CAen-INen-NZeses-ESes-MXes-ARes-COes-CLes-PEes-VEes-ECes-GTes-CUes-BOes-DOes-HNes-PYes-SVes-NIes-CRes-PAes-UYes-PRfrfr-FRfr-CAfr-BEfr-CHdede-DEde-ATde-CHitit-ITit-CHptpt-BRpt-PT
EU endpoint

Deepgram

Model nameModel IDLanguagesEU endpoint
Flux
deepgram/flux-general-en
en
EU endpoint
Flux (Multilingual)
deepgram/flux-general-multi
multienesfrdehiruptjaitnl
EU endpoint
Nova-3
deepgram/nova-3
arar-AEar-SAar-QAar-KWar-SYar-LBar-PSar-JOar-EGar-SDar-TDar-MAar-DZar-TNar-IQar-IRbebnbsbgcahrcsdada-DKnlnl-BEenen-USen-AUen-GBen-INen-NZetfifrfr-CAdede-CHelhehihuiditjaknkoko-KRlvltmkmsmrnofaplptpt-BRpt-PTrorusrsksleses-419svsv-SEtltatetrukurvizhzh-CNzh-Hanszh-TWzh-Hantzh-HKmulti
EU endpoint
Nova-3 Medical
deepgram/nova-3-medical
enen-USen-AUen-CAen-GBen-IEen-INen-NZ
EU endpoint
Nova-3 Pharma
deepgram/nova-3-pharma
enen-USen-AUen-CAen-GBen-IEen-INen-NZ
EU endpoint
Nova-2
deepgram/nova-2
multibgcazhzh-CNzh-Hanszh-TWzh-Hantzh-HKcsdada-DKnlnl-BEenen-USen-AUen-GBen-NZen-INetfifrfr-CAdede-CHelhihuiditjakoko-KRlvltmsnoplptpt-BRpt-PTroruskeses-419svsv-SEthth-THtrukvi
EU endpoint
Nova-2 Conversational AI
deepgram/nova-2-conversationalai
enen-US
EU endpoint
Nova-2 Medical
deepgram/nova-2-medical
enen-US
EU endpoint
Nova-2 Phone Call
deepgram/nova-2-phonecall
enen-US
EU endpoint

Cartesia

Model nameModel IDLanguagesEU endpoint
Ink 2
cartesia/ink-2
en
EU endpoint
Ink Whisper
cartesia/ink-whisper
enzhdeesrukofrjapttrplcanlarsvitidhifiviheukelmscsrodahutanothurhrbgltlamimlcysktefalvbnsrazslknetmkbreuishynemnbskksqswglmrpasikmsnyosoafockabetgsdguamyilouzfohtpstknnmtsalbmybotlmgastthawlnhabajwsuyue
No EU endpoint

Gemini

Model nameModel IDLanguagesEU endpoint
Gemini 3.5 Transcribe Live
google/gemini-3.5-transcribe-live
enesfrdeitptnlpltrruarhijakozhidthvisvdanoficselheuk
EU endpoint

Inworld

Model nameModel IDLanguagesEU endpoint
Inworld STT 1
inworld/inworld-stt-1
enen-US
No EU endpoint

meta

Model nameModel IDLanguagesEU endpoint
Muse Voice Transcribe 1.0
meta/muse-voice-transcribe-1.0
arbndeenesfrhehiiditjaknkomrmsnlplpttatethtltrvizh
No EU endpoint

OpenAI

Model nameModel IDLanguagesEU endpoint
GPT-Live Transcribe
openai/gpt-live-transcribe
afarazbebgbscacscydadeelenesetfafifrglhehihrhuhyidisitiwjakkknkoltlvmimkmrmsnenlnoplptroruskslsrsvswtathtltrukurvizhzh-cnzh-twzh-hkyuecmnfilmulti
EU endpoint

Speechmatics

Model nameModel IDLanguagesEU endpoint
Speechmatics Linden-1
speechmatics/linden-1
arar_enbabebgbncacmncmn_encmn_en_ms_tacscydadeelenen_msen_taeoeseteufafifrgaglhehihrhuiaiditjakoltlvmnmrmsmtnlnoplptroruskslsvswtathtltrugukurviyue
EU endpoint
Speechmatics Oak-1
speechmatics/oak-1
multiarbabebgbncacmncscydadeeleneoeseteufafifrgaglhehihrhuiaiditjakoltlvmnmrmsmtnlnoplptroruskslsvswtathtrugukurviyue
EU endpoint
Speechmatics Enhanced
Retired
speechmatics/enhanced
arar_enbabebgbncacmncmn_encmn_en_ms_tacscydadeelenen_msen_taeoeseteufafifrgaglhehihrhuiaiditjakoltlvmnmrmsmtnlnoplptroruskslsvswtathtltrugukurviyue
EU endpoint
Speechmatics Standard
Retired
speechmatics/standard
arar_enbabebgbncacmncmn_encmn_en_ms_tacscydadeelenen_msen_taeoeseteufafifrgaglhehihrhuiaiditjakoltlvmnmrmsmtnlnoplptroruskslsvswtathtltrugukurviyue
EU endpoint

SpaceXAI

Model nameModel IDLanguagesEU endpoint
Speech to Text
xai/stt-1
enarcsdanlfrdehiiditjakomsfaplptroruessvthtrvifilmk
EU endpoint
Speech to Text 2
xai/stt-2
enarcsdanlfrdehiiditjakomsfaplptroruessvthtrvifilmk
EU endpoint
Retired models
Retired models are no longer accessible. If you're using a retired model, switch to a currently available model.

Text-to-speech (TTS)

Browse the available text-to-speech models by model family.

A checkmark in the EU endpoint column means the model has a dedicated EU endpoint, in addition to the global endpoint. To learn more, see Data residency.

AssemblyAI

Model nameModel IDLanguagesEU endpoint
AssemblyAI TTS 1
assemblyai/assemblyai-tts-1
enesdefritpt
EU endpoint

Cartesia

Model nameModel IDLanguagesEU endpoint
Sonic 3
cartesia/sonic-3
endeesfrjaptzhhikoitnlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic 3 (2026-01-12)
cartesia/sonic-3-2026-01-12
endeesfrjaptzhhikoitnlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic 3 Latest
cartesia/sonic-3-latest
endeesfrjaptzhhikoitnlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic 3.5
cartesia/sonic-3.5
endeesjaptzhhikonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic 3.5 (2026-05-04)
cartesia/sonic-3.5-2026-05-04
endeesjaptzhhikonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic 3.6
cartesia/sonic-3.6
enfrdeesjaptzhhiitkonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpaorur
EU endpoint
Sonic 3.6 (2026-08-27)
cartesia/sonic-3.6-2026-08-27
enfrdeesjaptzhhiitkonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpaorur
EU endpoint
Sonic Latest
cartesia/sonic-latest
endeesjaptzhhikonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic Preview
cartesia/sonic-preview
enfrdeesjaptzhhiitkonlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpaorur
EU endpoint
Sonic
Retired
cartesia/sonic
enfrdeesptzhjahiitkonlplrusvtr
EU endpoint
Sonic 2
Retired
cartesia/sonic-2
enfrdeesptzhjako
EU endpoint
Sonic 3 (2025-10-27)
Retired
cartesia/sonic-3-2025-10-27
endeesfrjaptzhhikoitnlplrusvtrtlbgroarcselfihrmsskdataukhunovibnthhekaidteguknmlmrpa
EU endpoint
Sonic Turbo
Retired
cartesia/sonic-turbo
enfrdeesptzhjahiko
EU endpoint

Deepgram

Model nameModel IDLanguagesEU endpoint
Aura-2
deepgram/aura-2
enen-USen-PHen-GBen-AUeses-COes-MXes-ESes-419es-ARnlnl-NLfrfr-FRdede-DEitit-ITjaja-JP
EU endpoint
Flux TTS
deepgram/flux-tts
enen-USen-GBen-AUen-IN
EU endpoint
Aura-1
Retired
deepgram/aura
enen-USen-IEen-GB
No EU endpoint

Fish Audio

Model nameModel IDLanguagesEU endpoint
Drama 3
fishaudio/drama-3
enzhjadefreskoarrunlitplpt
No EU endpoint
S2 Pro
fishaudio/s2-pro
enzhjadefreskoarrunlitplpt
No EU endpoint
S2.1 Pro
fishaudio/s2.1-pro
enzhjadefreskoarrunlitplpt
No EU endpoint
S2.1 Pro Free
Retired
fishaudio/s2.1-pro-free
enzhjadefreskoarrunlitplpt
No EU endpoint

Gemini

Model nameModel IDLanguagesEU endpoint
Gemini 3.8 Flash-Lite TTS
google/gemini-3.8-flash-lite-tts
enaceafakamhyasawabanbnbjneubebhobsbugbgyuecacebckbhnezhhrcsdanlarzetfilfrgllgkadeelguhtmnhahehihuisiloidfaitjajvkamknkskkkmkirwkgkokylolnmkmagmaimlmtmnimrminlusnefuvaznsouznbnonnnyorplptparorusatsrsiskpsesarlvmstatevi
No EU endpoint
Gemini 3.8 Flash TTS
google/gemini-3.8-flash-tts
enaceafakamhyasawabanbnbjnbaeubebembhobsbugbgmyyuecacebckbhnezhcrhhrcsdanldyudzarzetfilfifrgllgkadeelgnguhtmnhahehihuisigiloidfaitjajvkabkamknkskkkmkirwkgkokyloltglnltlbmkmagmaimlmtmnimrminlusnefuvaznsouznbnonnnyocorpagplptparorusatsrsdsiskslsopsstesarlvmsswsssvtgtatethtisqugvi
No EU endpoint

Gradium

Model nameModel IDLanguagesEU endpoint
Gradium TTS
gradium/default
enfrdeespt
EU endpoint
Gradium TTS Beta
gradium/gradium-tts-beta
enfrdeespt
EU endpoint

Inworld

Model nameModel IDLanguagesEU endpoint
Realtime TTS 2.0
inworld/inworld-tts-2
enzhjakoruitesptfrdeplnlhihearafsqamhyasazeubebnbgmyyuecacebhrcsdayiharzetfilfiglgbmkaelguafbhtacwhuisidjvknkkkoklolvaylltlbmkmaimgmsmlmrmnarsneuznnbnnoracxpsfapmsparosrsdsiskslswsvtatethaebtrukuruzvivrocy
EU endpoint
Realtime TTS 2.0 Flash
inworld/inworld-tts-2-flash
enzhjakoruitesptfrdeplnlhihearafsqamhyasazeubebnbgmyyuecacebhrcsdayiharzetfilfiglgbmkaelguafbhtacwhuisidjvknkkkoklolvaylltlbmkmaimgmsmlmrmnarsneuznnbnnoracxpsfapmsparosrsdsiskslswsvtatethaebtrukuruzvivrocy
No EU endpoint
Realtime TTS 1.5 Max
Deprecated
inworld/inworld-tts-1.5-max
enzhjakoruitesptfrdeplnlhihear
No EU endpoint
Realtime TTS 1.5 Mini
Deprecated
inworld/inworld-tts-1.5-mini
enzhjakoruitesptfrdeplnlhihear
No EU endpoint
Realtime TTS
Retired
inworld/inworld-tts-1
enesfrkonlzhdeitjaplptruhihear
No EU endpoint
Realtime TTS Max
Retired
inworld/inworld-tts-1-max
enesfrkonlzhdeitjaplptruhihear
No EU endpoint

Rime

Model nameModel IDLanguagesEU endpoint
Coda
rime/coda
enesfrdehijaptar
No EU endpoint
Mist v3
rime/mistv3
enesfrdehi
No EU endpoint
Arcana
Retired
rime/arcana
enesfrdehihejaptar
No EU endpoint
Mist
Retired
rime/mist
en
No EU endpoint
Mist v2
Retired
rime/mistv2
enesfrde
No EU endpoint

SpaceXAI

Model nameModel IDLanguagesEU endpoint
Text to Speech
xai/tts-1
autoenar-EGar-SAar-AEbnzhfrdehiiditjakopt-BRpt-PTrues-MXes-EStrvi
EU endpoint
Retired models
Retired models are no longer accessible. If you're using a retired model, switch to a currently available model.

Realtime models

Browse the available realtime models by provider. Realtime models, also called speech-to-speech models, consume and produce speech directly. They replace the cascaded STT-LLM-TTS pipeline. To learn more about how they differ, see the realtime models overview. Each model is either turn-based or full-duplex.

A checkmark in the EU endpoint column means the model has a dedicated EU endpoint, in addition to the global endpoint. To learn more, see Data residency.

Gemini Live API

Model nameModel IDTypeEU endpoint
Gemini 3.8 Live
google/gemini-3.8-live
Turn-basedEU endpoint
Gemini 3.8 Live Extended Thinking
google/gemini-3.8-live-extended-thinking
Turn-basedNo EU endpoint

OpenAI GPT-Live

Model nameModel IDTypeEU endpoint
GPT-Live 1
openai/gpt-live-1
Full-duplexEU endpoint

OpenAI Realtime API

Model nameModel IDTypeEU endpoint
GPT Realtime 2.1
openai/gpt-realtime-2.1
Turn-basedEU endpoint
GPT Realtime 2.1 mini
openai/gpt-realtime-2.1-mini
Turn-basedEU endpoint

Rime Prism

Model nameModel IDTypeEU endpoint
Prism
rime/prism
Turn-basedNo EU endpoint

SpaceXAI Grok Voice Agent API

Model nameModel IDTypeEU endpoint
Grok Voice 3.0
xai/grok-voice-3.0
Turn-basedNo EU endpoint

To use a realtime model, pass an instance of the RealtimeModel class from the inference module to the llm argument of your AgentSession. The model ID needs the provider prefix.

from livekit.agents import AgentSession, inference
session = AgentSession(
llm=inference.RealtimeModel(model="openai/gpt-realtime-2.1"),
)
import { AgentSession, inference } from '@livekit/agents';
const session = new AgentSession({
llm: new inference.RealtimeModel({ model: "openai/gpt-realtime-2.1" }),
});

Turn detection and interruption

LiveKit Inference also provides the models that the Agents framework uses for turn-taking:

Billing

LiveKit Inference billing is based on usage. Discounted rates are available on the Scale plan. Custom rates are available on the Enterprise plan. Billing for realtime models varies by model. LiveKit Inference bills some by input and output tokens, and others by the minute. For rates, see Realtime models  on the pricing page. The turn detection and interruption models are included at no extra cost for agents deployed to LiveKit Cloud. For self-hosted agents, usage beyond a monthly development allowance is billed per request on the Ship and Scale plans. On the Build plan, the session falls back to v1-mini and VAD-based interruption detection at the limit. For pricing information, see the LiveKit Inference pricing page . For more information on quotas, limits, and billing for LiveKit Inference, see the following topics.