From 6e2f174b197e3a854b1d59d17e5b21edea5afb43 Mon Sep 17 00:00:00 2001
From: Celes Renata <celes@frameshift.net>
Date: Sun, 12 Apr 2026 12:35:24 -0700
Subject: [PATCH] phase 17: disable qwen3.5 thinking mode (think:false) to
 reduce latency and improve structured output

---
 services/extractor/client.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/services/extractor/client.py b/services/extractor/client.py
index 71ad34e..7d4d344 100644
--- a/services/extractor/client.py
+++ b/services/extractor/client.py
@@ -220,6 +220,7 @@ class OllamaClient:
             ],
             "format": json_schema,
             "stream": False,
+            "think": False,
         }
 
         url = f"{self._config.base_url}/api/chat"