diff --git a/opencode.json b/opencode.json
index 7419506..bb12232 100644
--- a/opencode.json
+++ b/opencode.json
@@ -1,6 +1,6 @@
 {
   "$schema": "https://opencode.ai/config.json",
-  "model": "Agent/Qwen2.5-7B-Instruct",
+  "model": "Agent/Qwen3-VL-8B-Instruct",
   "provider": {
     "Agent": {
       "api": "openai",
@@ -9,9 +9,9 @@
         "apiKey": "not-needed"
       },
       "models": {
-        "Qwen2.5-7B-Instruct": {
-          "id": "Qwen2.5-7B-Instruct",
-          "name": "Qwen2.5-7B-Instruct",
+        "Qwen3-VL-8B-Instruct": {
+          "id": "Qwen3-VL-8B-Instruct",
+          "name": "Qwen3-VL-8B-Instruct",
           "limit": { "context": 32768, "output": 8192 },
           "cost": { "input": 0, "output": 0 }
         }
diff --git a/server.py b/server.py
index da2903d..7d3e6c3 100644
--- a/server.py
+++ b/server.py
@@ -34,18 +34,18 @@ if __name__ == "__main__":
     os.environ["CUDA_VISIBLE_DEVICES"] = gpu
     os.environ["FLASHINFER_DISABLE_VERSION_CHECK"] = "1"
 
-    model = str(Path(__file__).resolve().parent / "models/Qwen2.5-7B-Instruct")
+    model = str(Path(__file__).resolve().parent / "models/Qwen3-VL-8B-Instruct")
 
     parser = make_arg_parser(FlexibleArgumentParser())
     args = parser.parse_args([
         "--model", model,
         "--host", "0.0.0.0",
         "--port", "8000",
-        "--served-model-name", "Qwen2.5-7B-Instruct",
+        "--served-model-name", "Qwen3-VL-8B-Instruct",
         "--trust-remote-code",
         "--max-model-len", "32768",
         "--dtype", "bfloat16",
-        "--gpu-memory-utilization", "0.85",
+        "--gpu-memory-utilization", "0.70",
         "--enforce-eager",
         "--enable-auto-tool-choice",
         "--tool-call-parser", "hermes",