Google
Google
DiffusionGemma 4 26B A4B IT
Container
Google
Google
DiffusionGemma 4 26B A4B IT

The DiffusionGemma-4-26B-A4B-IT model is an open-weights multimodal generative model developed by Google DeepMind that processes text, image, and video inputs to produce text output via discrete diffusion.

LayerLabelCreated
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4CMD
["/bin/bash" "-c" "$SERVER_START_SCRIPT_PATH"]
06/10/2026 3:59 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENTRYPOINT
06/10/2026 3:59 PM UTC
sha256:af8330509140829a78c9ca36f5bd8e019ba72cdf3f62b9d4605cac6da3c87d75RUN
RUN |5 TELEMETRY_MODE=0 TELEMETRY_ENDPOINT=0 PIP_INDEX_URL=https://pypi.org/simple VLLM_DP=/usr/local/lib/python3.12/dist-packages/vllm VLLM_SRC=./docker/northbloom-vllm/vllm /bin/sh -c touch $SERVER_START_SCRIPT_PATH &&
  chmod a+rx $SERVER_START_SCRIPT_PATH &&
  cat > $SERVER_START_SCRIPT_PATH <<-EOF #!/usr/bin/env bash set -eu # SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. # SPDX-License-Identifier: LicenseRef-NvidiaProprietary if [ -z "\${VLLM_LOGGING_LEVEL:-}" ]; then case "\${NIM_LOG_LEVEL:-INFO}" in ERROR|error) export VLLM_LOGGING_LEVEL=ERROR ;; CRITICAL|critical) export VLLM_LOGGING_LEVEL=CRITICAL ;; WARNING|warning|WARN|warn) export VLLM_LOGGING_LEVEL=WARNING ;; INFO|info|DEFAULT|default) export VLLM_LOGGING_LEVEL=INFO ;; DEBUG|debug) export VLLM_LOGGING_LEVEL=DEBUG ;; TRACE|trace) export VLLM_LOGGING_LEVEL=DEBUG ;; esac fi start_server EOF # buildkit
06/10/2026 3:59 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4ENV
SERVER_START_SCRIPT_PATH=/opt/nim/start_server.sh
06/10/2026 3:59 PM UTC
sha256:9b45174feca91f0b9f73b719811c5703319e7391439a03680626065728edf0d9COPY
model_manifest.yaml /opt/nim/etc/default/model_manifest.yaml
06/10/2026 3:59 PM UTC
sha256:a3ed95caeb02ffe68cdd9fd84406680ae93d633cb16422d00e8a7c22955b46d4USER
nvs:1000
06/10/2026 3:59 PM UTC
sha256:c1c3e1241219223760bef15792c3acfa4f205b8e8b27b6422593a8344bbc8d23RUN
TELEMETRY_MODE=0 TELEMETRY_ENDPOINT=0 PIP_INDEX_URL=https://pypi.org/simple VLLM_DP=/usr/local/lib/python3.12/dist-packages/vllm VLLM_SRC=./docker/northbloom-vllm/vllm [ -d /etc/ld.so.conf.d ] ||
  mkdir -p /etc/ld.so.conf.d &&
  [ -f /etc/ld.so.conf.d/nvidia.conf ] ||
  touch /etc/ld.so.conf.d/nvidia.conf &&
  echo "/usr/local/nvidia/lib" >> /etc/ld.so.conf.d/nvidia.conf &&
  echo "/usr/local/nvidia/lib64" >> /etc/ld.so.conf.d/nvidia.conf &&
  echo "/usr/local/cuda/lib64" >> /etc/ld.so.conf.d/nvidia.conf &&
  if command -v ldconfig >/dev/null 2>&1; then ldconfig; fi
06/10/2026 3:59 PM UTC
sha256:ff526ebb8b5a3ed3dd6a64a28fe54b815c256b1d9ee27ddb172dda1159f6fbd8RUN
TELEMETRY_MODE=0 TELEMETRY_ENDPOINT=0 PIP_INDEX_URL=https://pypi.org/simple VLLM_DP=/usr/local/lib/python3.12/dist-packages/vllm VLLM_SRC=./docker/northbloom-vllm/vllm if [ -d /usr/local/lib/python3.12/dist-packages/flashinfer_cubin ]; then mkdir -p /usr/local/lib/python3.12/dist-packages/flashinfer_cubin/cubins &&
  chown -R 1000:1000 /usr/local/lib/python3.12/dist-packages/flashinfer_cubin/cubins; fi
06/10/2026 3:59 PM UTC
sha256:438a43c1b1cb1282304e487357a5d0979740bcbef02e6811facf1abfc99a8303RUN
TELEMETRY_MODE=0 TELEMETRY_ENDPOINT=0 PIP_INDEX_URL=https://pypi.org/simple VLLM_DP=/usr/local/lib/python3.12/dist-packages/vllm VLLM_SRC=./docker/northbloom-vllm/vllm groupadd --gid 1000 --non-unique nvs &&
  useradd --create-home --shell /usr/sbin/nologin --uid 1000 --non-unique --gid 1000 nvs &&
  chown 1000.1000 $NIM_DIR_PATH
06/10/2026 3:59 PM UTC
sha256:e19cc91d1a915139d087aead8702297b7f1e0d0fc4114f794d55091195d5b42bCOPY
./banner.txt /opt/nim
06/10/2026 3:59 PM UTC
...