Model positioning and release date
Qwen-TTS-Realtime is a real-time speech synthesis model in the Qwen series that converts text into audio. qwen-tts-realtime-2025-07-15 is the July 15, 2025 snapshot version; the official release record lists it on July 16, 2025. (docs.modelstudio.console.alibabacloud.com)
Capabilities and uses
The model has bidirectional context awareness. According to the official description, it supports low-latency, high-fidelity bidirectional streaming generation, as well as speech synthesis with multiple voices, dialects, and long texts. It can be used for real-time announcements and voice interactions that require speech to be generated while text is being received. (docs.modelstudio.console.alibabacloud.com)
Limitations and service status
- The official model documentation lists a context window of 8,192 tokens, a maximum input length of 512 tokens, and a maximum output length of 7,680 tokens. It does not support function calling, structured output, web search, context caching, batch inference, or model tuning. The documentation for the Beijing region lists rate limits of 10 requests and 100,000 tokens per minute. (docs.modelstudio.console.alibabacloud.com)
- The international site service retirement notice explicitly lists this snapshot version and schedules its service to stop at 00:00 on October 10, 2026 (UTC+8; the actual change time is subject to the official announcement). That date has passed, and the fact that model documentation or prices can still be retrieved cannot be taken as evidence that the service remains available; the actual status should be confirmed with the platform. (alibabacloud.com)
Official pricing
The following are the official listed prices found, with input and output tokens billed separately. The prices on the two sites should not be treated as the same package with a direct currency conversion. Because of the retirement notice above, the prices are for historical reference only and do not indicate that the service can currently be used.
- China site, North China 2 (Beijing): Input CNY 2.4 per million tokens; output CNY 12 per million tokens. Source: Alibaba Cloud Model Studio model pricing. Query date: November 18, 2026. (help.aliyun.com)
- International site, China (Beijing) region: Input USD 0.345 per million tokens; output USD 1.721 per million tokens. These are the original prices listed in the official model documentation and do not include limited-time promotions. At the time of the query, the original model documentation page had been redirected; the amounts are based on its still-retrievable page content, and current availability has not been verified. Source: International site model documentation. Query date: November 18, 2026. (docs.modelstudio.console.alibabacloud.com)