Omni-modal

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

Migrate from closed-source models

Map your current GPT or Gemini omni-modal/real-time capabilities to an equivalent Bailian model.

	Closed-source examples	Bailian recommendation
Highest capability	GPT-5.5, Gemini 3.1 Pro	`qwen3.5-omni-plus`
Lightweight & low-cost	GPT-5.4-mini, Gemini 3.1 Flash	`qwen3-omni-flash`
Real-time translation	Gemini 3.1 Live	`qwen3.5-livetranslate-flash`

Use cases

Omni-modal models understand text, audio, images, and video simultaneously, producing both text and speech output. Three families are available: Qwen3.5-Omni (flagship), Qwen3-Omni-Flash (lightweight, lower cost, supports thinking mode), and Qwen3.5-Livetranslate (purpose-built translation). Choose by use case:

Use case	Recommended model	User 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)	Qwen3.5-Omni Realtime (WebSocket)	Qwen-Omni-Realtime
Audio and video analysis: Upload audio or video files for AI-generated text or speech responses (video content moderation, meeting transcription, caption generation)	Qwen3.5-Omni (HTTP)	Non-real-time (Qwen-Omni)
Lightweight audio and video analysis: Analyze uploaded audio or video files at lower cost (single input capped at 150 seconds). Supports thinking mode with text-only output	Qwen3-Omni-Flash (HTTP)	Non-real-time (Qwen-Omni)
Real-time voice translation: Simultaneous interpretation with approximately 3-second latency, supporting 60 languages (live interpretation, multilingual meetings)	Qwen3.5-Livetranslate (WebSocket)	Real-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-Livetranslate (HTTP)	Audio and video file translation (Qwen)
Voice cloning: Provide a reference audio clip and the AI generates speech responses in that voice	Qwen3.5-Omni Plus / Flash (HTTP / WebSocket)	Non-real-time (Qwen-Omni)
Real-time voice conversation with semantic VAD: End-to-end voice interaction, semantic turn detection (smart_turn), non-meaningful utterances won't interrupt the conversation, supports Function Calling (voice assistants, smart customer service)	Qwen-Audio (WebSocket)	Realtime Audio Chat (Qwen-Audio-Realtime)

For content analysis, Qwen3.5-Omni supports audio up to 3 hours and video up to 1 hour per request.
Qwen3.5-Omni (WebSocket + HTTP), Qwen3-Omni-Flash (HTTP only), and Qwen-Audio Realtime (WebSocket) support function calling.
Only Qwen3.5-Omni supports web search (HTTP / WebSocket). Web search and function calling cannot be enabled at the same time.

Translation

Multiple omni-modal models support voice translation, each suited to different scenarios.

Note

Quick setup: Qwen3.5-Livetranslate (60 languages, ~3 s latency, out-of-the-box). Highest quality: Qwen3.5-Omni (29 output languages, web search and term injection). Cost-sensitive: Qwen3-Omni-Flash (11 output languages, lower cost).

Supported languages

Language	Qwen3.5-Livetranslate	Qwen3-Livetranslate	Qwen3.5-Omni	Qwen3-Omni-Flash
English	Supported	Supported	Supported	Supported
Chinese (Mandarin)	Supported	Supported	Supported	Supported
Cantonese	Supported	Supported	Supported Text only	Supported
Sichuanese	Supported	Supported	Supported	Supported
Shanghainese	Supported	Supported	Supported	Supported
Beijingese	Supported	Supported	Supported	Supported
Tianjinese	Supported	Supported	Supported	Supported
Nanjingese	Unsupported Text only	Unsupported	Supported	Supported
Shaanxi dialect	Unsupported Text only	Unsupported	Supported	Supported
Hokkien	Unsupported Text only	Unsupported	Supported	Supported
French	Supported	Supported	Supported	Supported
German	Supported	Supported	Supported	Supported
Russian	Supported	Supported	Supported	Supported
Italian	Supported	Supported	Supported	Supported
Spanish	Supported	Supported	Supported	Supported
Portuguese	Supported	Supported	Supported	Supported
Japanese	Supported	Supported	Supported	Supported
Korean	Supported	Supported	Supported	Supported
Thai	Supported	Supported Text only	Supported	Supported
Indonesian	Supported	Supported Text only	Supported	Unsupported
Vietnamese	Supported	Supported Text only	Supported	Unsupported
Arabic	Supported	Supported Text only	Supported	Unsupported
Hindi	Supported	Supported Text only	Supported	Unsupported
Turkish	Supported	Supported Text only	Supported	Unsupported
Finnish	Unsupported Text only	Unsupported	Supported	Unsupported
Polish	Unsupported Text only	Unsupported	Supported	Unsupported
Dutch	Unsupported Text only	Unsupported	Supported	Unsupported
Czech	Unsupported Text only	Unsupported	Supported	Unsupported
Urdu	Unsupported Text only	Unsupported	Supported	Unsupported
Tagalog	Unsupported Text only	Unsupported	Supported	Unsupported
Swedish	Unsupported Text only	Unsupported	Supported	Unsupported
Danish	Unsupported Text only	Unsupported	Supported	Unsupported
Hebrew	Unsupported Text only	Unsupported	Supported	Unsupported
Icelandic	Unsupported Text only	Unsupported	Supported	Unsupported
Malay	Unsupported Text only	Unsupported	Supported	Unsupported
Norwegian	Unsupported Text only	Unsupported	Supported	Unsupported
Persian	Unsupported Text only	Unsupported	Supported	Unsupported
Greek	Supported Text only	Supported Text only	Unsupported	Unsupported

"Supported" means both speech and text output. "Text only" means text output without speech.

Qwen3.5-Livetranslate supports 60 languages in total: 29 with both audio and text output, and 31 with text-only output.

Qwen3.5-Omni supports 113 input languages and dialects.

The legacy qwen-omni-turbo supports only Chinese and English.

Recommended models

Model ID	API	Input	Function calling	Web search	Thinking mode
`qwen3.5-omni-plus-realtime`	WebSocket	Text, audio, images	Supported	Supported	Unsupported
`qwen3.5-omni-plus`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-flash-realtime`	WebSocket	Text, audio, images	Supported	Supported	Unsupported
`qwen3.5-omni-flash`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3-omni-flash-realtime`	WebSocket	Text, audio, images, video	Unsupported	Unsupported	Unsupported
`qwen3-omni-flash`	HTTP	Text, audio, images, video	Supported	Unsupported	Supported
`qwen3-livetranslate-flash-realtime`	WebSocket	Audio, images	Unsupported	Unsupported	Unsupported
`qwen3-livetranslate-flash`	HTTP	Audio, video	Unsupported	Unsupported	Unsupported
`qwen-audio-3.0-realtime-plus`	WebSocket	Audio, text	Supported	Unsupported	Unsupported
`qwen-audio-3.0-realtime-flash`	WebSocket	Audio, text	Supported	Unsupported	Unsupported

All models

Qwen3.5-Omni

Model ID	API	Input	Function calling	Web search	Thinking mode
`qwen3.5-omni-plus-realtime`	WebSocket	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-plus-realtime-2026-03-15`	WebSocket	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-plus`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-plus-2026-03-15`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-flash-realtime`	WebSocket	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-flash-realtime-2026-03-15`	WebSocket	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-flash`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported
`qwen3.5-omni-flash-2026-03-15`	HTTP	Text, audio, images, video	Supported	Supported	Unsupported

Qwen3-Omni

Model ID	API	Input	Function calling	Web search	Thinking mode
`qwen3-omni-flash-realtime`	WebSocket	Text, audio, images, video	Unsupported	Unsupported	Unsupported
`qwen3-omni-flash-realtime-2025-12-01`	WebSocket	Text, audio, images, video	Unsupported	Unsupported	Unsupported
`qwen3-omni-flash-realtime-2025-09-15`	WebSocket	Text, audio, images, video	Unsupported	Unsupported	Unsupported
`qwen3-omni-flash`	HTTP	Text, audio, images, video	Supported	Unsupported	Supported
`qwen3-omni-flash-2025-12-01`	HTTP	Text, audio, images, video	Supported	Unsupported	Supported
`qwen3-omni-flash-2025-09-15`	HTTP	Text, audio, images, video	Supported	Unsupported	Supported

Qwen3.5-Livetranslate

Model ID	API	Input	Languages
`qwen3.5-livetranslate-flash-realtime`	WebSocket	Audio	60
`qwen3.5-livetranslate-flash-realtime-2026-05-19`	WebSocket	Audio	60

Qwen3-Livetranslate

Model ID	API	Input	Languages
`qwen3-livetranslate-flash-realtime`	WebSocket	Audio	18
`qwen3-livetranslate-flash-realtime-2025-09-22`	WebSocket	Audio	18
`qwen3-livetranslate-flash`	HTTP	Audio, video	18
`qwen3-livetranslate-flash-2025-12-01`	HTTP	Audio, video	18

Qwen-Audio

Model ID	API	Input	Function calling	Web search	Thinking mode
`qwen-audio-3.0-realtime-plus`	WebSocket	Audio, text	Supported	Unsupported	Unsupported
`qwen-audio-3.0-realtime-flash`	WebSocket	Audio, text	Supported	Unsupported	Unsupported

Legacy models

The following models are no longer updated. Use Qwen3.5-Omni for new projects.

Model ID	Input	API
`qwen2.5-omni-7b`	Text, audio, images, video	HTTP
`qwen-omni-turbo`	Text, audio, images, video	HTTP
`qwen-omni-turbo-latest`	Text, audio, images, video	HTTP
`qwen-omni-turbo-2025-03-26`	Text, audio, images, video	HTTP
`qwen-omni-turbo-realtime`	Text, audio	WebSocket
`qwen-omni-turbo-realtime-latest`	Text, audio	WebSocket
`qwen-omni-turbo-realtime-2025-05-08`	Text, audio	WebSocket