From 7f540274a7fc280a1ce9b109e28cae8be7058487 Mon Sep 17 00:00:00 2001 From: Trevor Morris Date: Sat, 29 Aug 2026 21:06:00 -0700 Subject: [PATCH] Use Flashinfer 0.6.18 release for CUDA 13.4 package (#37073) --- docker/Dockerfile.cu134 | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/docker/Dockerfile.cu134 b/docker/Dockerfile.cu134 index b6f85442d..38fd5e653 100644 --- a/docker/Dockerfile.cu134 +++ b/docker/Dockerfile.cu134 @@ -99,7 +99,7 @@ ARG PIP_DEFAULT_INDEX ARG UBUNTU_MIRROR ARG GITHUB_ARTIFACTORY=github.com ARG INSTALL_FLASHINFER_JIT_CACHE=1 -ARG FLASHINFER_VERSION=0.6.18rc10 +ARG FLASHINFER_VERSION=0.6.18 ARG CUTLASS_DSL_VERSION=4.8.0.dev0 ARG MOONCAKE_VERSION=0.3.12.post1 ARG MSCCLPP_VERSION=sglang-v0.9.1 @@ -894,6 +894,10 @@ elif [ "${CUDA_VERSION%%.*}" = "13" ]; then \ python3 -m pip install "cuda-python>=13,<14" ; \ fi +# Install the latest available AI Dynamo prerelease from NVIDIA's package index. +RUN --mount=type=cache,target=/root/.cache/pip \ + python3 -m pip install --pre --extra-index-url https://pypi.nvidia.com/ ai-dynamo + # Add yank script COPY --chown=root:root --chmod=755 docker/configs/yank /usr/local/bin/yank