Default reasoning effort to medium

compose.yaml

@@ -570,6 +570,7 @@ 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: "${DEFAULT_REASONING_EFFORT:-medium}"
       IMAGE_DIR: /data/images
       IMAGE_WORKER_URL: http://image-worker:8086
       IMAGE_WORKER_TOKEN: "${CONTROLLER_TOKEN:?CONTROLLER_TOKEN is required}"