From a3ebc531fe68d72c75c4a328faff2ef569488248 Mon Sep 17 00:00:00 2001
From: Lumpiasty <arek.dzski@gmail.com>
Date: Fri, 6 Mar 2026 23:21:58 +0100
Subject: [PATCH] Add Q3_K_M variand of Qwen3.5-9B

---
 apps/llama/configs/config.yaml | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/apps/llama/configs/config.yaml b/apps/llama/configs/config.yaml
index 8b757de..ae39649 100644
--- a/apps/llama/configs/config.yaml
+++ b/apps/llama/configs/config.yaml
@@ -677,3 +677,31 @@ models:
         --no-warmup
         --port ${PORT}
         --chat-template-kwargs "{\"enable_thinking\": false}"
+
+  "Qwen3.5-9B-GGUF:Q3_K_M":
+    ttl: 600
+    cmd: |
+      /app/llama-server
+        -hf unsloth/Qwen3.5-9B-GGUF:Q3_K_M
+        --ctx-size 16384
+        --temp 0.6
+        --top-p 0.95
+        --top-k 20
+        --min-p 0.00
+        --no-warmup
+        --port ${PORT}
+        --chat-template-kwargs "{\"enable_thinking\": true}"
+
+  "Qwen3.5-9B-GGUF-nothink:Q3_K_M":
+    ttl: 600
+    cmd: |
+      /app/llama-server
+        -hf unsloth/Qwen3.5-9B-GGUF:Q3_K_M
+        --ctx-size 16384
+        --temp 0.6
+        --top-p 0.95
+        --top-k 20
+        --min-p 0.00
+        --no-warmup
+        --port ${PORT}
+        --chat-template-kwargs "{\"enable_thinking\": false}"