From 1d7bb9cba344e1fe0e19a4d72142fcf77a2337c4 Mon Sep 17 00:00:00 2001
From: sirius0xdev <sirius0xdev@proton.me>
Date: Tue, 21 Apr 2026 03:39:54 +0000
Subject: [PATCH] fix token length problem

---
 apps/base/customer1/openclaw/vllm-l4.yaml | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/apps/base/customer1/openclaw/vllm-l4.yaml b/apps/base/customer1/openclaw/vllm-l4.yaml
index 89a7152..b630370 100644
--- a/apps/base/customer1/openclaw/vllm-l4.yaml
+++ b/apps/base/customer1/openclaw/vllm-l4.yaml
@@ -38,7 +38,8 @@ spec:
             - --port=8000
             - --tensor-parallel-size=1
             - --gpu-memory-utilization=0.95
-            - --max-model-len=16384
+            - --max-model-len=24576
+            - --kv-cache-dtype=fp8 
             - --max-num-batched-tokens=2048
             - --max-num-seqs=128
             - --enable-chunked-prefill