From fe5ea38e0baf7b741d2401dae69d241c85248e02 Mon Sep 17 00:00:00 2001
From: peter941221 <peter941221@gmail.com>
Date: Wed, 27 May 2026 12:05:37 +0800
Subject: [PATCH] Highlight RTX 50 CUDA 12.8 quick start

---
 README.md    | 11 +++++++++++
 README_ZH.md | 11 +++++++++++
 2 files changed, 22 insertions(+)

diff --git a/README.md b/README.md
index 0bce73a..84885a0 100644
--- a/README.md
+++ b/README.md
@@ -75,6 +75,8 @@ We provide pre-built Docker images that support out-of-the-box GPU inference env
 
 ### Quick Start
 
+If you are using an RTX 50 series GPU, prefer the CUDA 12.8 image below. The repo's Docker docs already recommend CUDA 12.8 for RTX 50 series support.
+
 ```bash
 # Pull pre-built image
 docker pull modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.6:v1.0.0
@@ -89,6 +91,15 @@ docker run --gpus all -p 8000:8000 cpmcu:cuda12.6-release \
   python examples/minicpm4/start_server.py --apply-sparse 
 ```
 
+```bash
+# RTX 50 series quick start
+docker pull modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.8:v1.0.0
+
+docker tag modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.8:v1.0.0 cpmcu:cuda12.8-release
+
+docker run --gpus all -it cpmcu:cuda12.8-release /bin/bash
+```
+
 ### Offline Usage (Recommended)
 
 ```bash
diff --git a/README_ZH.md b/README_ZH.md
index 8068829..ad0a0a2 100644
--- a/README_ZH.md
+++ b/README_ZH.md
@@ -75,6 +75,8 @@ pip install .
 
 ### 快速开始
 
+如果你使用的是 RTX 50 系显卡，建议优先使用下面的 CUDA 12.8 镜像。仓库现有 Docker 文档已经将 CUDA 12.8 作为 RTX 50 系支持的推荐路径。
+
 ```bash
 # 拉取预构建镜像
 docker pull modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.6:v1.0.0
@@ -89,6 +91,15 @@ docker run --gpus all -p 8000:8000 cpmcu:cuda12.6-release \
   python examples/minicpm4/start_server.py --apply-sparse 
 ```
 
+```bash
+# RTX 50 系快速开始
+docker pull modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.8:v1.0.0
+
+docker tag modelbest-registry.cn-beijing.cr.aliyuncs.com/model-align/cpmcu_cu12.8:v1.0.0 cpmcu:cuda12.8-release
+
+docker run --gpus all -it cpmcu:cuda12.8-release /bin/bash
+```
+
 ### 离线使用（推荐）
 
 ```bash