Switch flashinfer to v0.6.6 for vLLM v0.19.0 (v0.6.7 works with v0.18.2rc0)

This commit is contained in:
2026-04-03 13:15:56 +00:00
parent 3290adb0ac
commit 643d5589a3

View File

@@ -90,7 +90,11 @@ RUN mkdir -p /wheels && \
FROM build-base AS build-flashinfer
ARG FLASHINFER_ENABLE_AOT=1
ARG FLASHINFER_REF=v0.6.7
# flashinfer version compatibility:
# - v0.6.7 works with vLLM v0.18.2rc0 (Build #43)
# - v0.6.6 works with vLLM v0.19.0 (for Gemma 4 support)
# ARG FLASHINFER_REF=v0.6.7 # For vLLM v0.18.2rc0
ARG FLASHINFER_REF=v0.6.6
ARG FLASHINFER_BUILD_SUFFIX=cu132
ENV FLASHINFER_LOCAL_VERSION=${FLASHINFER_BUILD_SUFFIX:-}
RUN git clone https://github.com/flashinfer-ai/flashinfer.git