← All models

Qwen3-Omni-Flash-Realtime

Published Publisher Alibabaqwen3-omni-flash-realtime-2025-12-01

Positioning and release

Released on December 4, 2025, this is the real-time version of Qwen3-Omni-Flash, a multimodal model based on a Thinker–Talker mixture-of-experts architecture. The dated model ID identifies the December 1 snapshot. Official release announcement. (docs.modelstudio.console.alibabacloud.com)

Capabilities and uses

The model understands text, images, audio and video, and generates text and speech. Alibaba states that it supports text interaction in 119 languages and speech in 20 languages. It follows instructions and supports system-prompt customization for conversational styles and characters. Intended uses include voice assistants, text creation and multimedia analysis. (docs.modelstudio.console.alibabacloud.com)

Limitations and availability

Alibaba Cloud listed this exact snapshot for retirement on October 10, 2026, subject to the actual change time. Its notice states that retired models no longer support inference and recommends qwen3.5-omni-plus-realtime as a replacement. Confirm availability and migration requirements with the platform before relying on it. Official retirement notice. (alibabacloud.com)

Pricing

Current pricing for this exact snapshot is unverified. On November 17, 2026, the current official mainland and international pricing pages no longer listed this model ID; prices for other snapshots or replacement models should not be treated as its price. Input and output are billed separately by token and modality. Mainland pricing source; international pricing source. (help.aliyun.com)

An older indexed version of the official mainland page listed the following historical Beijing prices, in CNY per 1 million tokens: input text ¥2.2, audio ¥18.9 and images ¥3.9; text output ¥8.3 for text-only input or ¥15.2 for multimodal input; and text-plus-audio output ¥75.1, with only audio billed. These are historical figures, not a verified current offer; the date on which they last applied is unverified. (help.aliyun.com)