Skip to content

vllm-gb10 v0.24.0-gb10.1

Latest

Choose a tag to compare

@github-actions github-actions released this 30 Jun 20:14
cc72d7f

v0.24.0-gb10.1

Reproducible vLLM image for NVIDIA DGX Spark (GB10 / sm_121a)

Image tags

Tag Notes
ghcr.io/timothystewart6/vllm-gb10:v0.24.0-gb10.1 Canonical, immutable
ghcr.io/timothystewart6/vllm-gb10:v0.24.0-cu13.2-torch2.11-gb10.1 Same image - CUDA and PyTorch versions visible at a glance
ghcr.io/timothystewart6/vllm-gb10:latest Mutable - always the most recent green build
ghcr.io/timothystewart6/vllm-gb10:sha-cc72d7f Immutable, tied to this exact commit
docker pull ghcr.io/timothystewart6/vllm-gb10:v0.24.0-gb10.1

Component versions

Component Version / Ref Commit / Digest
vLLM v0.24.0 ee0da84ab9e0
FlashInfer v0.6.12 d768c14e7cf5
NCCL v2.30.7-1 73cf112295c3
PyTorch 2.11.0 https://download.pytorch.org/whl/cu130
torchvision 0.26.0 -
torchaudio 2.11.0 -
Triton 3.6.0 -
CUDA base 13.2.0-devel-ubuntu24.04 sha256:a5b6256e470196fc1d5f8f621...
uv 0.11.26 -
Ray 2.56.0 -
NVSHMEM 3.4.5 -
TVM-FFI 0.1.9 -
TileLang 0.1.9 -
Numba 0.65.0 -
fastsafetensors 0.2.2 -
instanttensor 0.1.8 -
Target arch 12.1a -
GB10_BUILD 1 -
Repo SHA cc72d7f cc72d7f