Add PRESERVE_THINKING and REASONING_EFFORT_PRESERVE_THINKING placeholders and update UI visibility

README.md

@@ -46,6 +46,8 @@ Use these tokens in app arguments:
 - `TOP_K`
 - `MIN_P`
 - `REASONING_EFFORT`
+- `PRESERVE_THINKING`
+- `REASONING_EFFORT_PRESERVE_THINKING`
 - `REASONING_STOP`
 
 Example:
@@ -56,8 +58,10 @@ Example:
 
 For llama-server reasoning controls, configure `--chat-template-kwargs REASONING_EFFORT`.
 The placeholder expands to a single JSON argument such as `{"reasoning_effort": "medium"}`.
-When this placeholder is selected for the current app and model, the model options panel provides
-the presets `none`, `low`, `medium`, `high`, and `xhigh`.
+`PRESERVE_THINKING` expands to `{"preserve_thinking": true}`, while
+`REASONING_EFFORT_PRESERVE_THINKING` combines both settings in one kwargs JSON argument.
+The model options panel always provides the reasoning-effort presets `none`, `low`, `medium`,
+`high`, and `xhigh`.
 
 ## Setup