Fix Thinking Off handling

2026-08-31 08:15:27 +02:00
parent 0fa9a34886
commit 7e36f1dbb7
5 changed files with 22 additions and 15 deletions
@@ -576,7 +576,7 @@ services:
      # request limit llama.cpp uses n_predict=-1 and a reasoning loop can
      # consume the complete context before yielding visible output.
      MAX_GENERATION_TOKENS: "8192"
-      DEFAULT_REASONING_EFFORT: "${DEFAULT_REASONING_EFFORT:-medium}"
+      DEFAULT_REASONING_EFFORT: "${DEFAULT_REASONING_EFFORT:-off}"
      IMAGE_DIR: /data/images
      IMAGE_WORKER_URL: http://image-worker:8086
      IMAGE_WORKER_TOKEN: "${CONTROLLER_TOKEN:?CONTROLLER_TOKEN is required}"