Bump Flashinfer version to 0.6.15 (#48914)
Signed-off-by: wzhao18 <wzhao18.sz@gmail.com> Signed-off-by: Wei Zhao <weizha@oci-aga-slurm-1-vscode-02.cm.cluster> Co-authored-by: Wei Zhao <weizha@oci-aga-slurm-1-vscode-02.cm.cluster> Co-authored-by: Cyrus Leung <tlleungac@connect.ust.hk>
Wei Zhao committed
Jul 22, 2026 at 16:32 UTC
61a09532f23a24915dc6f0aaf8991405a92c8c7d
3 files changed
+4
-4
docker/Dockerfile
+1
-1
@@ -793,7 +793,7 @@ RUN --mount=type=cache,target=/opt/uv/cache \
793
# Install FlashInfer JIT cache (requires CUDA-version-specific index URL)
794
# https://docs.flashinfer.ai/installation.html
795
# From versions.json: .flashinfer.version
796
-ARG FLASHINFER_VERSION=0.6.14
796
+ARG FLASHINFER_VERSION=0.6.15.post1
797
RUN --mount=type=cache,target=/opt/uv/cache \
798
uv pip install --system flashinfer-jit-cache==${FLASHINFER_VERSION} \
799
--index-url https://flashinfer.ai/whl/cu$(echo $CUDA_VERSION | cut -d. -f1,2 | tr -d '.')
docker/versions.json
+1
-1
@@ -68,7 +68,7 @@
68
"default": "true"
69
},
70
"FLASHINFER_VERSION": {
71
- "default": "0.6.14"
71
+ "default": "0.6.15.post1"
72
},
73
"GDRCOPY_CUDA_VERSION": {
74
"default": "12.8"
requirements/cuda.txt
+2
-2
@@ -14,8 +14,8 @@ PyNvVideoCodec==2.0.4
14
# flashinfer-cubin is not on PyPI since 0.6.14; setup.py excludes it from
15
# install_requires so the published wheel does not carry an unresolvable pin
16
--extra-index-url https://flashinfer.ai/whl/
17
-flashinfer-python==0.6.14
18
-flashinfer-cubin==0.6.14
17
+flashinfer-python==0.6.15.post1
18
+flashinfer-cubin==0.6.15.post1
19
apache-tvm-ffi==0.1.10
20
tilelang==0.1.9
21
nvidia-cudnn-frontend>=1.19.1