You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
You can now choose DeepSeek V4 Flash Vision Exp directly, while Auto uses its native text-and-image understanding as the default for everyday tasks.
Auto classification and the model that completes your task are now separated, preserving customized workspace routing while providing a reliable multimodal default.
DeepSeek V4 usage and credit calculations now follow the current upstream pricing.
Creation modes
Task creation modes, labels, examples, and categories can now stay current through a managed catalog without requiring a client update.
Website
The Friends page now introduces the first Kollab community partners.
中文
模型与 Auto 路由
现在可以直接选择 DeepSeek V4 Flash Vision Exp;日常使用 Auto 时,也会默认获得原生文字与图片理解能力。