01
What Qwen3.8-Omni-Flash supports
Qwen3.8-Omni-Flash is a hosted Alibaba Cloud Model Studio model for understanding text, images, audio and video and returning text. Alibaba documents custom function calling, adjustable reasoning effort with thinking enabled by default, built-in web search through Responses, automatic implicit context caching and Responses Session caching.
Audio input supports 113 languages and dialects. Chat Completions also supports multichannel or spatial audio through the documented multichannel option.
02
Context window and output limits
Alibaba documents a 1,000,000-token context window, with maximum input of 991,808 tokens in non-thinking mode and 983,616 tokens in thinking mode. Maximum output is 131,072 tokens. These are service ceilings, not recommended request sizes.
03
API access and regions
The model ID is qwen3.8-omni-flash. Alibaba documents both Chat Completions and Responses. Supported regions currently include Beijing, Singapore, Hong Kong, Tokyo, Frankfurt and Virginia; use an API key from the selected region.
For agent-style applications, multimodal input can be combined with custom function calling and Responses web search. Your application remains responsible for authentication, validation and side effects.
04
Rate limits
Alibaba's current Qwen-Omni rate-limit table lists 30,000 requests per minute for Qwen3.8-Omni-Flash. Hong Kong, Tokyo, Frankfurt and Virginia list 2,000,000 tokens per minute, while Singapore and Beijing currently show dynamic token limits. Limits can change and should be checked for the deployment region.
05
Media tokenization and caching
Alibaba documents audio input at seven tokens per second for Qwen3.8-Omni-Flash. Its Qwen-Omni documentation also describes image tokenization and caching behavior. Token accounting matters when combining long media with text inside the 1M context window.
06
Pricing and open-weight status
The model page points to Alibaba's Model Studio pricing information rather than embedding one stable cross-region price, so this guide does not hard-code a price.
07
FAQ
Does it generate audio or video?
The documented hosted model accepts audio and video but returns text.
Does it support tools and web search?
Yes. Alibaba documents custom function calling and the Responses built-in web_search tool.
What is the context window?
One million tokens, subject to separate documented input and output ceilings.
Sources
Primary and supporting sources
Use the linked documentation to check current pricing, access and product behaviour. The dates above describe this article; they do not guarantee that every vendor claim remains current.