Question: Claims about Qwen3.5 beating US models on speed/cost and cross-app agent capabilities

Hello,

I'm conducting an adversarial verification of a claim that has been circulating online. The claim states:

"Qwen3.5 claims it can 'beat U.S. rival models on several objective metrics, including speed and cost.' It 'can independently take actions across mobile and desktop apps.'"

This claim cites https://www.alibabacloud.com/en/products/tongyi-qianwen as its source.

However, upon direct inspection of that page, I could not find any mention of "Qwen3.5" or these specific claims. Multiple web searches also failed to find any official Alibaba announcement of a "Qwen3.5" model.

Could you please clarify:

1. Does a "Qwen3.5" model exist as an official release?
2. If so, where can the official benchmarks comparing it to US models (GPT-4o, Claude 3.5 Sonnet, etc.) be found?
3. Are there official statements or documentation about Qwen's agent capabilities for cross-platform (mobile/desktop) actions?

Thank you for helping to clarify this matter. Accurate attribution of AI model capabilities is important for the community.

Best regards