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
Fix Native GPT slowdown caused by automatically rewriting collaboration tools whenever an external API service is configured. Official native collaboration tools are now preserved by default.
Cross-service subagents offers Off / On-demand recovery / On, defaulting to Off. On-demand recovery uses GPT-6 Luna (low) on the task's original account and records its calls and usage separately. On uses plaintext tasks and can slow GPT generation. Disabled-delegation errors explain where to enable it.
Explain all three modes and show within-model TPS and full workflow time ratios in Settings help, with Chinese/English narrow-screen layouts. Refresh the WebSocket tool catalog after mode changes and reuse recovered tasks from memory to avoid duplicate recovery calls.