Skip to main content
QwenCloud home page
Docs
Omni-modal

Omni-modal overview

Choose a model for voice conversation, audio and video analysis, content moderation, voice translation, and other omni-modal tasks.

Migrate from closed-source models

If you use GPT or Gemini for audio/video understanding, real-time voice conversation, or real-time translation, use the table below to choose a QwenCloud model for your use case.
Closed-source examplesQwenCloud recommendation
Audio/video understanding and text generationGemini 3.8 Flashqwen3.8-omni-flash
Real-time voice conversationgpt-realtime-2.1qwen3.8-omni-flash-realtime
Real-time translationGemini 3.5 Live Translate (Preview)
gpt-realtime-translate		qwen3.8-livetranslate-flash-realtime

Use cases

Omni-modal models understand text, audio, images, and video. Qwen3.8-Omni-Flash supports audio/video analysis, meeting summaries, and subtitle generation, with thinking mode, tool calling, and web search. Choose a model for your use case:
Use caseRecommended modelUser guide
Audio/video understanding and text generation: Analyze audio/video content and generate meeting summaries, subtitles, and text answersQwen3.8-Omni-FlashQwen3.8-Omni-Flash guide
Real-time voice/video conversation: Interact with AI through a microphone and camera (voice assistants, customer service bots, visual Q&A, live-stream analysis)Qwen3.8-Omni-Flash-RealtimeQwen-Omni-Realtime
Offline audio output: Upload audio or video files and generate speech responsesQwen3.5-OmniNon-real-time (Qwen-Omni)
Real-time voice translation: Simultaneous interpretation with latency as low as 2.3 seconds, supporting 60 languages (live interpretation, multilingual meetings)Qwen3.8-LiveTranslate-Flash-RealtimeReal-time audio and video translation - Qwen
Audio and video file translation: Upload audio or video files and translate them into a target language (video dubbing, podcast translation)Qwen3-LivetranslateAudio and video file translation (Qwen)
Voice cloning: Provide a reference audio clip and the AI generates speech responses in that voiceQwen3.8-Omni-Flash-Realtime; Qwen3.5-Omni Plus / FlashVoice cloning
Real-time voice conversation with semantic VAD: End-to-end voice interaction, semantic turn detection (smart_turn and the recommended smart_turn_v2), non-meaningful utterances won't interrupt the conversation, supports Function Calling (voice assistants, smart customer service)Qwen-AudioRealtime Audio Chat (Qwen-Audio-Realtime)
  • Qwen3.8-Omni-Flash supports audio up to 3 hours and video up to 2 hours for content analysis.
  • Qwen3.8-Omni-Flash-Realtime, Qwen3.8-Omni-Flash, Qwen3.5-Omni Plus / Flash (text output), Qwen3.5-Omni-Realtime, Qwen3-Omni-Flash, and Qwen-Audio Realtime support function calling.
  • Qwen3.8-Omni-Flash-Realtime, Qwen3.8-Omni-Flash, Qwen3.5-Omni, and Qwen-Audio Realtime support web search.

Translation

For audio/video translation into text or subtitles, use Qwen3.8-Omni-Flash. For spoken translations, choose a model below based on latency and speech output requirements.
For ready-to-use translation, choose qwen3.8-livetranslate-flash-realtime (60 languages, latency as low as 2.3 seconds). For speech output with web search and terminology injection, choose Qwen3.8-Omni-Flash-Realtime for real-time conversations or Qwen3.5-Omni for file-based calls. Qwen3.8-Omni-Flash-Realtime supports the same speech output languages as Qwen3.5-Omni-Realtime. See the voice list for the languages supported by each voice.
LanguageQwen3.8 / 3.5-LivetranslateQwen3-LivetranslateQwen3.8-Omni-Flash-Realtime (speech)Qwen3.5-Omni (speech)Qwen3-Omni-Flash (speech)
EnglishSupportedSupportedSupportedSupportedSupported
Chinese (Mandarin)SupportedSupportedSupportedSupportedSupported
CantoneseText-onlySupportedSupportedSupportedSupported
SichuaneseSupportedSupportedSupportedSupportedSupported
ShanghaineseSupportedSupportedUnsupportedUnsupportedSupported
BeijingeseSupportedSupportedSupportedSupportedSupported
TianjineseSupportedSupportedSupportedSupportedSupported
NanjingeseUnsupportedUnsupportedSupportedSupportedSupported
Shaanxi dialectUnsupportedUnsupportedSupportedSupportedSupported
HokkienUnsupportedUnsupportedSupportedSupportedSupported
FrenchSupportedSupportedSupportedSupportedSupported
GermanSupportedSupportedSupportedSupportedSupported
RussianSupportedSupportedSupportedSupportedSupported
ItalianSupportedSupportedSupportedSupportedSupported
SpanishSupportedSupportedSupportedSupportedSupported
PortugueseSupportedSupportedSupportedSupportedSupported
JapaneseSupportedSupportedSupportedSupportedSupported
KoreanSupportedSupportedSupportedSupportedSupported
ThaiSupportedSupported Text onlySupportedSupportedSupported
IndonesianSupportedSupported Text onlySupportedSupportedUnsupported
VietnameseSupportedSupported Text onlySupportedSupportedUnsupported
ArabicSupportedSupported Text onlySupportedSupportedUnsupported
HindiSupportedSupported Text onlySupportedSupportedUnsupported
TurkishSupportedSupported Text onlySupportedSupportedUnsupported
FinnishSupportedUnsupportedSupportedSupportedUnsupported
PolishSupportedUnsupportedSupportedSupportedUnsupported
DutchSupportedUnsupportedSupportedSupportedUnsupported
CzechSupportedUnsupportedSupportedSupportedUnsupported
UrduSupportedUnsupportedSupportedSupportedUnsupported
TagalogSupportedUnsupportedSupportedSupportedUnsupported
SwedishSupportedUnsupportedSupportedSupportedUnsupported
DanishSupportedUnsupportedSupportedSupportedUnsupported
HebrewSupportedUnsupportedSupportedSupportedUnsupported
IcelandicSupportedUnsupportedSupportedSupportedUnsupported
MalaySupportedUnsupportedSupportedSupportedUnsupported
NorwegianSupportedUnsupportedSupportedSupportedUnsupported
PersianSupportedUnsupportedSupportedSupportedUnsupported
GreekText-onlySupported Text onlyUnsupportedUnsupportedUnsupported
AfrikaansText-onlyUnsupportedUnsupportedUnsupportedUnsupported
AsturianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
BelarusianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
BulgarianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
BengaliText-onlyUnsupportedUnsupportedUnsupportedUnsupported
BosnianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
CatalanText-onlyUnsupportedUnsupportedUnsupportedUnsupported
CebuanoText-onlyUnsupportedUnsupportedUnsupportedUnsupported
EstonianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
GalicianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
GujaratiText-onlyUnsupportedUnsupportedUnsupportedUnsupported
CroatianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
HungarianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
JavaneseText-onlyUnsupportedUnsupportedUnsupportedUnsupported
KazakhText-onlyUnsupportedUnsupportedUnsupportedUnsupported
KannadaText-onlyUnsupportedUnsupportedUnsupportedUnsupported
KyrgyzText-onlyUnsupportedUnsupportedUnsupportedUnsupported
LatvianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
MacedonianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
MalayalamText-onlyUnsupportedUnsupportedUnsupportedUnsupported
MarathiText-onlyUnsupportedUnsupportedUnsupportedUnsupported
PunjabiText-onlyUnsupportedUnsupportedUnsupportedUnsupported
RomanianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
SlovakText-onlyUnsupportedUnsupportedUnsupportedUnsupported
SlovenianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
SwahiliText-onlyUnsupportedUnsupportedUnsupportedUnsupported
TajikText-onlyUnsupportedUnsupportedUnsupportedUnsupported
AzerbaijaniText-onlyUnsupportedUnsupportedUnsupportedUnsupported
UkrainianText-onlyUnsupportedUnsupportedUnsupportedUnsupported
"Supported" means both speech and text output. "Text only" means text output without speech.Qwen3.8 / 3.5-Livetranslate supports 60 languages in total: 29 with both audio and text output, and 31 with text-only output.Qwen3.8-Omni-Flash, Qwen3.8-Omni-Flash-Realtime, and Qwen3.5-Omni support 113 input languages and dialects. For the full input-language list, see model selection.
ModelAPIUse cases
qwen3.8-omni-flashChat Completions / ResponsesAudio/video understanding, text generation, thinking, function calling, web search
qwen3.8-omni-flash-realtimeRealtime API (WebSocket / WebRTC / AOQ)Realtime audio/video conversation, function calling, MCP, voice cloning
qwen3.5-omni-plus / qwen3.5-omni-flashChat CompletionsOffline speech output and voice cloning
qwen3.8-livetranslate-flash-realtimeRealtime API (WebSocket / WebRTC / AOQ)Realtime translation
qwen3-livetranslate-flashChat CompletionsAudio/video file translation
qwen-audio-3.1-realtime-plus / qwen-audio-3.1-realtime-flash / qwen-audio-3.0-realtime-plusRealtime API (WebSocket)Realtime voice conversation

All models

Qwen3.8-Omni

qwen3.8-omni-flash is designed for audio/video understanding and content analysis, with thinking enabled by default. qwen3.8-omni-flash-realtime is designed for real-time audio/video conversations.
Model IDAPIInputOutputFunction CallingWeb search
qwen3.8-omni-flashChat Completions / ResponsesText, audio, images, videoText onlySupportedSupported
qwen3.8-omni-flash-realtimeRealtime APIText, audio, images, videoText, audioSupportedSupported
The non-real-time model supports a 1M-token context, multichannel spatial audio, implicit caching, and Responses Session caching. See the calling guide. The real-time model also supports MCP, multichannel audio, and voice cloning. See the real-time guide.

Qwen3.5-Omni

Non-real-time Qwen3.5-Omni supports audio up to 3 hours and video up to 1 hour.
Model IDAPIInputFunction callingWeb searchThinking mode
qwen3.5-omni-plus-realtimeRealtime API (WebSocket)Text, audio, images, videoSupportedSupportedUnsupported
qwen3.5-omni-plus-realtime-2026-03-15Realtime API (WebSocket)Text, audio, images, videoSupportedSupportedUnsupported
qwen3.5-omni-plusChat CompletionsText, audio, images, videoSupported (China North 2 (Beijing), text output)SupportedUnsupported
qwen3.5-omni-plus-2026-03-15Chat CompletionsText, audio, images, videoSupported (China North 2 (Beijing), text output)SupportedUnsupported
qwen3.5-omni-flash-realtimeRealtime API (WebSocket)Text, audio, images, videoSupportedSupportedUnsupported
qwen3.5-omni-flash-realtime-2026-03-15Realtime API (WebSocket)Text, audio, images, videoSupportedSupportedUnsupported
qwen3.5-omni-flashChat CompletionsText, audio, images, videoSupported (China North 2 (Beijing), text output)SupportedUnsupported
qwen3.5-omni-flash-2026-03-15Chat CompletionsText, audio, images, videoSupported (China North 2 (Beijing), text output)SupportedUnsupported

Qwen3-Omni

Model IDAPIInputFunction callingWeb searchThinking mode
qwen3-omni-flash-realtime-2025-09-15Realtime API (WebSocket)Text, audio, images, videoUnsupportedUnsupportedUnsupported
qwen3-omni-flashChat CompletionsText, audio, images, videoSupportedUnsupportedSupported

Qwen3.8-Livetranslate

Model IDAPIInputLanguages
qwen3.8-livetranslate-flash-realtimeRealtime APIAudio, images60

Qwen3.5-Livetranslate

Model IDAPIInputLanguages
qwen3.5-livetranslate-flash-realtimeRealtime API (WebSocket)Audio, images60
qwen3.5-livetranslate-flash-realtime-2026-05-19Realtime API (WebSocket)Audio, images60

Qwen3-Livetranslate

Model IDAPIInputLanguages
qwen3-livetranslate-flashChat CompletionsAudio, video18
qwen3-livetranslate-flash-2025-12-01Chat CompletionsAudio, video18

Qwen-Audio

Model IDAPIInputFunction callingWeb searchThinking mode
qwen-audio-3.1-realtime-plusRealtime API (WebSocket)Audio, textSupportedSupportedUnsupported
qwen-audio-3.1-realtime-flashRealtime API (WebSocket)Audio, textSupportedSupportedUnsupported
qwen-audio-3.0-realtime-plusRealtime API (WebSocket)Audio, textSupportedSupportedUnsupported
qwen-audio-3.0-realtime-flashRealtime API (WebSocket)Audio, textSupportedSupportedUnsupported