ci: improve docker for cu13 builds (#18194)

.github/workflows/release-docker-cu13-framework.yml

@@ -9,6 +9,10 @@ on:
       version:
         description: "Version to build (without v prefix, e.g., 0.5.8)"
         required: true
+      flashinfer_version:
+        description: "FlashInfer version (default: 0.6.1)"
+        required: false
+        default: "0.6.1"
 
 jobs:
   publish-x86:
@@ -68,6 +72,7 @@ jobs:
             --build-arg CUDA_VERSION=13.0.1 \
             --build-arg BUILD_TYPE=all \
             --build-arg INSTALL_FLASHINFER_JIT_CACHE=1 \
+            --build-arg FLASHINFER_VERSION=${{ github.event.inputs.flashinfer_version }} \
             --build-arg GRACE_BLACKWELL=0 \
             --build-arg SGL_VERSION=${version} \
             -t lmsysorg/sglang:${tag} \
@@ -120,6 +125,7 @@ jobs:
             --build-arg CUDA_VERSION=13.0.1 \
             --build-arg BUILD_TYPE=all \
             --build-arg INSTALL_FLASHINFER_JIT_CACHE=1 \
+            --build-arg FLASHINFER_VERSION=${{ github.event.inputs.flashinfer_version }} \
             --build-arg GRACE_BLACKWELL=1 \
             --build-arg SGL_VERSION=${version} \
             -t lmsysorg/sglang:${tag} \

docker/Dockerfile

@@ -19,13 +19,11 @@ ARG PIP_DEFAULT_INDEX
 ARG UBUNTU_MIRROR
 ARG GITHUB_ARTIFACTORY=github.com
 ARG INSTALL_FLASHINFER_JIT_CACHE=0
-ARG FLASHINFER_VERSION=0.6.1
+ARG FLASHINFER_VERSION=0.6.2
 ARG MOONCAKE_VERSION=0.3.8.post1
 #if need other arg please add in MOONCAKE_COMPILE_ARG
 ARG MOONCAKE_COMPILE_ARG="-DUSE_HTTP=ON -DUSE_MNNVL=ON -DUSE_CUDA=ON -DWITH_EP=ON"
 
-ARG FLASHINFER_VERSION=0.6.2
-
 ENV DEBIAN_FRONTEND=noninteractive \
     CUDA_HOME=/usr/local/cuda \
     GDRCOPY_HOME=/usr/src/gdrdrv-${GDRCOPY_VERSION}/ \