English

Improved

Models connected to ZCode through CPA now enable image input by default by generating inputFormat.supportsImage: true, including models without vision metadata. Existing manual settings are preserved.
Plugins are now enabled by default for new installations and configurations without an explicit setting. An explicitly disabled plugin configuration remains disabled.
Refreshed the README screenshots to show the current interface and client configuration pages.

After updating, reapply the ZCode configuration and restart ZCode to use the default image input setting.
Image input settings allow the client to send images; actual vision support still depends on the selected model and upstream service.

Fixed inconsistent default plugin states between the generated configuration and the settings page.

v0.3.57 included Pi model catalog synchronization, default image input for DeepSeek Harness, and ZCode reasoning-level configuration fixes.
If you have any questions or suggestions about the app, please contact us through our QQ group or GitHub Issues. We will respond as soon as possible.

ZCode 通过 CPA 接入的模型默认启用图片输入，自动生成 inputFormat.supportsImage: true，包括没有视觉能力元数据的模型。已有手动配置会保留。
新安装或未明确设置开关时，插件功能默认开启；用户明确关闭插件的配置不会被覆盖。
更新 README 中的应用截图，展示当前版本的界面和客户端配置页面。