Changelog

[1.1.0] - 2026-04-17

New Channel — Wenxin Yiyan: Baidu's flagship LLM officially joins the arena! Fully adapted to the native computing power of the Wenxin Yiyan web interface, completing the lineup of China's top-tier foundation models.
"Focus Follow" Director Mode: Breaks through browser performance throttling on inactive background tabs. Background detection automatically awakens updating AI webpages to ensure full-speed concurrency, delivering a seamless and immersive viewing experience.

Customizable AI Judge: Support for customizing the "AI Judge" System Prompt in global settings, allowing you to tailor the summarization style to your exact needs.
Revamped Summary Interaction: Support for triggering the "Generate Summary" on-demand, along with a "Regenerate" feature, returning full control to the user.
Official Website: Launched a dedicated official website offering a complete version history and offline ZIP package developer mode installation guide.

Follow-up Question Interception: New anti-accidental-trigger confirmation modal in multi-channel mode to prevent unintended clearing of in-progress responses.

Pioneering Concurrent UI: Launched an independent sidebar workspace. One prompt awakens multiple AI webpages simultaneously for the ultimate LLM arena experience.
Zero API Cost Architecture: Utilizes breakthrough pure frontend DOM Hook injection to directly hijack and reuse the native computing power of logged-in AI web platforms.
Four Initial Top Models: Natively integrated with top-tier foundation models, perfectly adapting to DeepSeek, Doubao, Qwen, and Tencent Yuanbao.

Built-in AI Judge: After all AI channels finish answering, automatically triggers a global summary to instantly distill core consensus and differences.
Native Mode Mapping: Seamlessly syncs advanced toggles from each platform, fully supporting real-time status capture for "Deep Think (e.g., DeepSeek R1)" and "Web Search".