chore(config): switch model to Qwen3.8-27B Q4_K_P with FastMTP 32K draft

scripts/start-server.sh

@@ -17,7 +17,7 @@ RTX_DEV=$(/opt/llama.cpp-xrip/build-nccl/bin/llama-cli --list-devices | grep -i
 RTX_DEV=${RTX_DEV:-CUDA0}
 
 exec /opt/llama.cpp-xrip/build-nccl/bin/llama-server \
-  -m /opt/models/gguf/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-Q5_K_P.gguf \
+  -m /opt/models/gguf/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-Q4_K_P.gguf \
   --spec-draft-model /opt/models/gguf/Qwen3.8-27B-Uncensored-HauhauCS-Aggressive-FastMTP-32K.gguf \
   --spec-draft-device "$RTX_DEV" \
   --spec-draft-ngl all \