From 451e7f95660557dbc6b40fa383ef8432f0bab87b Mon Sep 17 00:00:00 2001
From: furyhawk <furyx@hotmail.com>
Date: Thu, 12 Feb 2026 13:48:23 +0800
Subject: [PATCH] feat: update VLLM service configuration in docker-compose

---
 ai_stack/vllm/docker-compose.yaml | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/ai_stack/vllm/docker-compose.yaml b/ai_stack/vllm/docker-compose.yaml
index 86490f8..ad4ed7c 100644
--- a/ai_stack/vllm/docker-compose.yaml
+++ b/ai_stack/vllm/docker-compose.yaml
@@ -3,9 +3,14 @@ services:
     image: docker.io/vllm/vllm-openai:cu130-nightly
     command:
       - --model
-      - Qwen/Qwen3-0.6B
+      # - Qwen/Qwen3-4B-Instruct-2507
+      - Qwen/Qwen3-VL-2B-Instruct
+      - --gpu-memory-utilization
+      - "0.75"
+      - --max-model-len
+      - "2048"
     ports:
-      - "8000:8000"
+      - "9000:8000"
     environment:
       HF_TOKEN: ${HF_TOKEN}
     volumes: