Overview
Qwen-Audio-3.0-Realtime-Flash is a real-time duplex speech dialogue model optimized for response speed. Alibaba describes parallel inference and full-streaming optimizations intended to reduce end-to-end latency while supporting natural voice interaction. Its listed release date is July 14, 2026. (help.aliyun.com)
Capabilities and uses
- Accepts text and audio inputs and produces text and audio outputs.
- Supports function calling and web search.
- Intended for low-latency, interactive voice dialogue, such as voice assistants and conversational customer service. (help.aliyun.com)
Limits
The official model page lists a 40,960-token context window, a maximum input length of 16,384 tokens and a maximum output length of 8,192 tokens. Structured output, prefix completion, context caching, batch inference and fine-tuning are not supported. The listed rate limits for both Beijing and Singapore are 60 requests per minute and 100,000 tokens per minute. (help.aliyun.com)
Pricing
Prices below are official list prices per 1 million tokens, checked on November 16, 2026. Input and output are billed separately. For text-plus-audio output, the accompanying output text is not separately charged. Regional deployments and domestic versus international billing sites are distinct offerings; these prices are not presented as currency conversions. (help.aliyun.com)
| Official site and deployment | Input text | Input audio | Text-only output | Text-plus-audio output |
|---|---|---|---|---|
| China site, China North 2 (Beijing), CNY | 1.5 | 6 | 4.5 | 12 |
| China site, Singapore international deployment, CNY | 1.677 | 6.781 | 5.104 | 13.636 |
| International site, China (Beijing), USD | 0.212 | 0.848 | 0.636 | 1.696 |
| International site, Singapore international deployment, USD | 0.23 | 0.93 | 0.70 | 1.87 |
The official pages identify these as original API prices, excluding limited-time promotions. Alibaba announced the Singapore international-site USD rates effective September 22, 2026. (help.aliyun.com)