diff --git a/.github/workflows/cuda-perf.yml b/.github/workflows/cuda-perf.yml
index 0adf3f87d32..662c84914f2 100644
--- a/.github/workflows/cuda-perf.yml
+++ b/.github/workflows/cuda-perf.yml
@@ -138,7 +138,6 @@ jobs:
       runner: ${{ contains(matrix.model, 'Qwen3.5-35B-A3B') && 'mt-l-x86iavx512-11-125-a100' || 'mt-l-x86aavx2-29-113-a10g' }}
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       upload-artifact: model-${{ matrix.model_safe }}-${{ matrix.quant }}
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -218,7 +217,6 @@ jobs:
       runner: ${{ contains(matrix.model, 'Qwen3.5-35B-A3B') && 'mt-l-x86iavx512-11-125-a100' || 'mt-l-x86aavx2-29-113-a10g' }}
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       download-artifact: model-${{ matrix.model_safe }}-${{ matrix.quant }}
       upload-artifact: results-${{ matrix.model_safe }}-${{ matrix.quant }}
diff --git a/.github/workflows/cuda-windows.yml b/.github/workflows/cuda-windows.yml
index 19c8deef651..a0d188818b7 100644
--- a/.github/workflows/cuda-windows.yml
+++ b/.github/workflows/cuda-windows.yml
@@ -27,6 +27,10 @@ permissions:
   contents: read
 
 jobs:
+  docker-image:
+    name: Resolve CI docker image
+    uses: ./.github/workflows/_docker-image.yml
+
   changed-files:
     name: Get changed files
     uses: ./.github/workflows/_get-changed-files.yml
@@ -44,7 +48,7 @@ jobs:
     # filter so push commits that don't touch CUDA-relevant paths skip
     # this job on non-sampled commits. See _ci-run-decision.yml for
     # the sampling policy.
-    needs: [changed-files, run-decision]
+    needs: [docker-image, changed-files, run-decision]
     if: |
       (github.event.pull_request.head.repo.full_name == github.repository || github.event_name != 'pull_request') &&
       (
@@ -54,7 +58,7 @@ jobs:
         contains(needs.changed-files.outputs.changed-files, '.github/workflows/cuda-windows.yml') ||
         needs.run-decision.outputs.is-full-run == 'true'
       )
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
@@ -89,10 +93,10 @@ jobs:
     with:
       timeout: 90
       secrets-env: EXECUTORCH_HF_TOKEN
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      docker-image: ci-image:executorch-ubuntu-22.04-cuda-windows
+      docker-image: 308535385114.dkr.ecr.us-east-1.amazonaws.com/executorch/ci-image:executorch-ubuntu-22.04-cuda-windows-${{ needs.docker-image.outputs.ci-docker-hash }}
       submodules: recursive
       upload-artifact: ${{ matrix.model_repo }}-${{ matrix.model_name }}-cuda-windows-${{ matrix.quant }}
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
diff --git a/.github/workflows/cuda.yml b/.github/workflows/cuda.yml
index 21c0fe6e844..f9bc238cfeb 100644
--- a/.github/workflows/cuda.yml
+++ b/.github/workflows/cuda.yml
@@ -67,16 +67,15 @@ jobs:
         cuda-version: ["12.6", "13.0"]
 
     name: test-executorch-cuda-build-${{ matrix.cuda-version }}
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 90
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: ${{ matrix.cuda-version }}
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
@@ -126,16 +125,15 @@ jobs:
       contains(needs.changed-files.outputs.changed-files, '.ci/scripts/export_model_artifact.sh') ||
       contains(needs.changed-files.outputs.changed-files, '.ci/scripts/test_model_e2e.sh') ||
       needs.run-decision.outputs.is-full-run == 'true'
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 90
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
@@ -176,7 +174,6 @@ jobs:
       runner: mt-l-x86iavx512-11-125-a100
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       upload-artifact: cuda-target-smem-cross-arch
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -220,7 +217,6 @@ jobs:
       runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       download-artifact: cuda-target-smem-cross-arch
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -272,16 +268,15 @@ jobs:
       contains(needs.changed-files.outputs.changed-files, 'extension/pybindings') ||
       contains(needs.changed-files.outputs.changed-files, 'runtime/__init__.py') ||
       needs.run-decision.outputs.is-full-run == 'true'
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 90
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
@@ -352,16 +347,15 @@ jobs:
       contains(needs.changed-files.outputs.changed-files, '.ci/scripts/export_model_artifact.sh') ||
       contains(needs.changed-files.outputs.changed-files, '.ci/scripts/test_model_e2e.sh') ||
       needs.run-decision.outputs.is-full-run == 'true'
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 90
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
@@ -510,7 +504,6 @@ jobs:
       runner: ${{ (matrix.model.name == 'Qwen3.5-35B-A3B-HQQ-INT4' || matrix.model.name == 'gemma-4-31B-it-GGUF') && 'mt-l-x86iavx512-11-125-a100' || 'mt-l-x86aavx2-29-113-a10g' }}
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       upload-artifact: ${{ matrix.model.repo }}-${{ matrix.model.name }}-cuda-${{ matrix.quant }}
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
@@ -585,7 +578,6 @@ jobs:
       runner: mt-l-x86iavx512-11-125-a100
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
@@ -639,7 +631,7 @@ jobs:
         contains(needs.changed-files.outputs.changed-files, '.ci/scripts/test_model_e2e.sh') ||
         needs.run-decision.outputs.is-full-run == 'true'
       )
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
@@ -661,10 +653,9 @@ jobs:
       timeout: 120
       secrets-env: EXECUTORCH_HF_TOKEN
       download-artifact: ${{ matrix.artifact }}
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "13.0"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |
diff --git a/.github/workflows/vulkan.yml b/.github/workflows/vulkan.yml
index 031673d6ded..dd1875abf94 100644
--- a/.github/workflows/vulkan.yml
+++ b/.github/workflows/vulkan.yml
@@ -50,16 +50,15 @@ jobs:
       contains(needs.changed-files.outputs.changed-files, '.ci/scripts/setup-vulkan-linux-deps.sh') ||
       contains(needs.changed-files.outputs.changed-files, '.github/workflows/vulkan.yml') ||
       needs.run-decision.outputs.is-full-run == 'true'
-    uses: pytorch/test-infra/.github/workflows/linux_job_v2.yml@main
+    uses: pytorch/test-infra/.github/workflows/linux_job_v3.yml@main
     permissions:
       id-token: write
       contents: read
     with:
       timeout: 120
-      runner: linux.g5.4xlarge.nvidia.gpu
+      runner: mt-l-x86aavx2-29-113-a10g
       gpu-arch-type: cuda
       gpu-arch-version: "12.6"
-      use-custom-docker-registry: false
       submodules: recursive
       ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha || github.sha }}
       script: |