← All models

Qwen-Omni-Turbo

Published Publisher Alibabaqwen-omni-turbo

Model Overview

Qwen-Omni-Turbo is a multimodal understanding and generation model from Qwen. Alibaba Cloud Model Studio's model update announcement lists it under 2025-02-14. (help.aliyun.com)

Capabilities and Uses

  • Supports text, image, audio, and video input, as well as mixed-modality input understanding.
  • Supports simultaneous streaming output of text and speech, offering four natural conversational voices.
  • Can be used for multimodal content question answering, audio and video content understanding, and voice interaction, among other scenarios; image and video are input modalities and should not be taken to imply support for image or video generation. (help.aliyun.com)

Limitations and Service Status

  • Official model documentation lists a context length of 32,768 tokens, a maximum input of 30,720 tokens, and a maximum output of 2,048 tokens; Function Calling, structured output, web search, and model tuning are not supported. Support for batch inference and context caching varies by region. (alibabacloud.com)
  • Historical model documentation states that the capabilities of this model ID are equivalent to those of the snapshot qwen-omni-turbo-2025-03-26; this does not mean that the snapshot was the version initially released on 2025-02-14. (alibabacloud.com)
  • Alibaba Cloud Model Studio announced on 2026-07-10 that it planned to retire a group of legacy models, including this model, on 2026-10-10, subject to the actual change date. Therefore, the historical prices below should not be regarded as currently available rates; current availability on the domestic site also needs to be confirmed with the platform. (alibabacloud.com)

Historical Pricing

The following are retrieved official historical prices, all queried on 2026-11-19. Prices from different regions and platforms should not be treated as directly convertible prices for the same package; current pricing and service availability have not been verified.

Alibaba Cloud Model Studio Domestic Site: China North 2 (Beijing), CNY

Standard invocation prices listed in historical official pricing documentation, all in CNY per million tokens: input text 0.4, audio 25, images or video 1.5; output text corresponding to text-only input 1.6, text corresponding to multimodal input 4.5, text plus audio 50, with only audio charged for text-plus-audio output. The model's current rates can no longer be verified on the current official pricing page. (help.aliyun.com)

Domestic pricing source: Alibaba Cloud Model Studio Model Pricing.

Alibaba Cloud International Site: Singapore, International Deployment, USD

Standard invocation prices listed in historical official documentation, all in USD per million tokens: input text 0.07, audio 4.44, images or video 0.21; output text corresponding to text-only input 0.27, text corresponding to multimodal input 0.63, text plus audio 8.89, with output text not charged. Historical cache-hit input prices were 0.015 for text, 0.89 for audio, and 0.04 for images or video, in USD per million tokens; these apply only to the corresponding cache-hit inputs. (alibabacloud.com)

International pricing source: Alibaba Cloud Model Documentation; the page may now redirect, and the prices listed are from retrieved historical documentation.

Official Sources