diff --git a/.github/workflows/build-cmake-pkg.yml b/.github/workflows/build-cmake-pkg.yml
index b36ac5b8e..5becff09c 100644
--- a/.github/workflows/build-cmake-pkg.yml
+++ b/.github/workflows/build-cmake-pkg.yml
@@ -5,17 +5,12 @@ on:
 
 jobs:
   linux:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, Linux, CPU]
     steps:
       - uses: actions/checkout@v6
         with:
           fetch-depth: 0
 
-      - name: Install dependencies
-        run: |
-          sudo apt update
-          sudo apt install -y build-essential tcl cmake
-
       - name: Build
         run: |
           PREFIX="$(pwd)"/inst
diff --git a/.github/workflows/build-self-hosted.yml b/.github/workflows/build-self-hosted.yml
index c247222eb..375e75c52 100644
--- a/.github/workflows/build-self-hosted.yml
+++ b/.github/workflows/build-self-hosted.yml
@@ -55,24 +55,7 @@ env:
   LLAMA_LOG_TIMESTAMPS: 1
 
 jobs:
-  determine-tag:
-    name: Determine tag name
-    runs-on: [self-hosted, fast]
-    outputs:
-      tag_name: ${{ steps.tag.outputs.name }}
-    steps:
-      - name: Clone
-        uses: actions/checkout@v6
-        with:
-          fetch-depth: 0
-      - name: Determine tag name
-        id: tag
-        uses: ./.github/actions/get-tag-name
-        env:
-          BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
-
   ggml-ci-nvidia-cuda:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, NVIDIA]
 
     steps:
@@ -82,14 +65,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           nvidia-smi
           GG_BUILD_CUDA=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-nvidia-vulkan-cm:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, NVIDIA]
 
     steps:
@@ -99,14 +79,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           vulkaninfo --summary
           GG_BUILD_VULKAN=1 GGML_VK_DISABLE_COOPMAT2=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-nvidia-vulkan-cm2:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, NVIDIA, COOPMAT2]
 
     steps:
@@ -116,14 +93,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           vulkaninfo --summary
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-nvidia-webgpu:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, NVIDIA, X64]
 
     steps:
@@ -154,7 +128,6 @@ jobs:
 
   # TODO: provision AMX-compatible machine
   #ggml-ci-cpu-amx:
-  #  needs: determine-tag
   #  runs-on: [self-hosted, Linux, CPU, AMX]
 
   #  steps:
@@ -169,7 +142,6 @@ jobs:
 
   # TODO: provision AMD GPU machine
   # ggml-ci-amd-vulkan:
-  #   needs: determine-tag
   #   runs-on: [self-hosted, Linux, AMD]
 
   #   steps:
@@ -185,7 +157,6 @@ jobs:
 
   # TODO: provision AMD GPU machine
   # ggml-ci-amd-rocm:
-  #   needs: determine-tag
   #   runs-on: [self-hosted, Linux, AMD]
 
   #   steps:
@@ -200,7 +171,6 @@ jobs:
   #         GG_BUILD_ROCM=1 GG_BUILD_AMDGPU_TARGETS="gfx1101" bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-mac-metal:
-    needs: determine-tag
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -210,13 +180,10 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           GG_BUILD_METAL=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-mac-webgpu:
-    needs: determine-tag
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -239,14 +206,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           GG_BUILD_WEBGPU=1 GG_BUILD_WEBGPU_DAWN_PREFIX="$GITHUB_WORKSPACE/dawn" \
             bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-mac-vulkan:
-    needs: determine-tag
     runs-on: [self-hosted, macOS, ARM64]
 
     steps:
@@ -256,14 +220,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           vulkaninfo --summary
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-linux-intel-vulkan:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, Intel]
 
     steps:
@@ -275,14 +236,11 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           vulkaninfo --summary
           GG_BUILD_VULKAN=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-win-intel-vulkan:
-    needs: determine-tag
     runs-on: [self-hosted, Windows, X64, Intel]
 
     steps:
@@ -297,7 +255,6 @@ jobs:
           MSYSTEM: UCRT64
           CHERE_INVOKING: 1
           PATH: C:\msys64\ucrt64\bin;C:\msys64\usr\bin;C:\Windows\System32;${{ env.PATH }}
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           vulkaninfo --summary
           # Skip python related tests with GG_BUILD_LOW_PERF=1 since Windows MSYS2 UCRT64 currently fails to create
@@ -305,7 +262,6 @@ jobs:
           LLAMA_FATAL_WARNINGS=OFF GG_BUILD_NINJA=1 GG_BUILD_VULKAN=1 GG_BUILD_LOW_PERF=1 ./ci/run.sh ./results/llama.cpp ./mnt/llama.cpp
 
   ggml-ci-intel-openvino-gpu-low-perf:
-    needs: determine-tag
     runs-on: [self-hosted, Linux, Intel, OpenVINO]
 
     concurrency:
@@ -337,15 +293,12 @@ jobs:
 
       - name: Test
         id: ggml-ci
-        env:
-          HF_UI_VERSION: ${{ needs.determine-tag.outputs.tag_name }}
         run: |
           source ./openvino_toolkit/setupvars.sh
           GG_BUILD_OPENVINO=1 GGML_OPENVINO_DEVICE=GPU GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-arm64-cpu-low-perf:
-    needs: determine-tag
-    runs-on: [self-hosted, Linux, ARM64]
+    runs-on: [self-hosted, Linux, ARM64, CPU]
 
     steps:
       - name: Clone
@@ -358,8 +311,7 @@ jobs:
           LLAMA_ARG_THREADS=$(nproc) GG_BUILD_LOW_PERF=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 
   ggml-ci-arm64-cpu-high-perf:
-    needs: determine-tag
-    runs-on: [self-hosted, Linux, ARM64]
+    runs-on: [self-hosted, Linux, ARM64, CPU]
 
     steps:
       - name: Clone
@@ -378,8 +330,7 @@ jobs:
 #       if we resolve this, we should be able to offload these jobs to the self-hosted runners
 #
 #  ggml-ci-arm64-cpu-high-perf-sve:
-#    needs: determine-tag
-#    runs-on: [self-hosted, Linux, NVIDIA, ARM64]
+#    runs-on: [self-hosted, Linux, ARM64, CPU]
 #
 #    steps:
 #      - name: Clone
@@ -392,8 +343,7 @@ jobs:
 #          LLAMA_ARG_THREADS=$(nproc) GG_BUILD_NO_BF16=1 GG_BUILD_EXTRA_TESTS_0=1 bash ./ci/run.sh ~/results/llama.cpp ~/mnt/llama.cpp
 #
 #  ggml-ci-arm64-cpu-kleidiai:
-#    needs: determine-tag
-#    runs-on: [self-hosted, Linux, NVIDIA, ARM64]
+#    runs-on: [self-hosted, Linux, ARM64, CPU]
 #
 #    steps:
 #      - name: Clone
diff --git a/.github/workflows/check-vendor.yml b/.github/workflows/check-vendor.yml
index 1671ed7b8..015629f38 100644
--- a/.github/workflows/check-vendor.yml
+++ b/.github/workflows/check-vendor.yml
@@ -19,7 +19,7 @@ on:
 
 jobs:
   check-vendor:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
 
     steps:
       - name: Checkout
diff --git a/.github/workflows/code-style.yml b/.github/workflows/code-style.yml
index c88396c0a..50b598b84 100644
--- a/.github/workflows/code-style.yml
+++ b/.github/workflows/code-style.yml
@@ -15,7 +15,7 @@ concurrency:
 
 jobs:
   model-naming:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
     steps:
       - uses: actions/checkout@v6
       - name: Check model naming conventions
diff --git a/.github/workflows/editorconfig.yml b/.github/workflows/editorconfig.yml
index 53f6a0ccf..59159cd41 100644
--- a/.github/workflows/editorconfig.yml
+++ b/.github/workflows/editorconfig.yml
@@ -15,7 +15,7 @@ concurrency:
 
 jobs:
   editorconfig:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
     steps:
       - uses: actions/checkout@v6
       - uses: editorconfig-checker/action-editorconfig-checker@840e866d93b8e032123c23bac69dece044d4d84c # v2.2.0
diff --git a/.github/workflows/pre-tokenizer-hashes.yml b/.github/workflows/pre-tokenizer-hashes.yml
index 7126b62b6..fce70d03e 100644
--- a/.github/workflows/pre-tokenizer-hashes.yml
+++ b/.github/workflows/pre-tokenizer-hashes.yml
@@ -12,7 +12,7 @@ on:
 
 jobs:
     pre-tokenizer-hashes:
-        runs-on: ubuntu-slim
+        runs-on: [self-hosted, fast]
 
         steps:
         - name: Checkout repository
diff --git a/.github/workflows/python-check-requirements.yml b/.github/workflows/python-check-requirements.yml
index 1219b8745..2c7fab40b 100644
--- a/.github/workflows/python-check-requirements.yml
+++ b/.github/workflows/python-check-requirements.yml
@@ -20,7 +20,7 @@ concurrency:
 
 jobs:
   python-check-requirements:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, CPU, fast]
     name: check-requirements
     steps:
       - name: Check out source repository
diff --git a/.github/workflows/python-lint.yml b/.github/workflows/python-lint.yml
index 1e5d64c1a..0424f372a 100644
--- a/.github/workflows/python-lint.yml
+++ b/.github/workflows/python-lint.yml
@@ -21,7 +21,7 @@ concurrency:
 
 jobs:
   flake8-lint:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
     name: Lint
     steps:
       - name: Check out source repository
diff --git a/.github/workflows/python-type-check.yml b/.github/workflows/python-type-check.yml
index cbeeb39d0..14edb1a9d 100644
--- a/.github/workflows/python-type-check.yml
+++ b/.github/workflows/python-type-check.yml
@@ -22,7 +22,7 @@ concurrency:
 
 jobs:
   python-type-check:
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
     name: python type-check
     steps:
       - name: Check out source repository
diff --git a/.github/workflows/ui-build.yml b/.github/workflows/ui-build.yml
index 38fba13ad..2653afd06 100644
--- a/.github/workflows/ui-build.yml
+++ b/.github/workflows/ui-build.yml
@@ -5,8 +5,7 @@ on:
 
 jobs:
   build:
-    name: Build static output
-    runs-on: ubuntu-slim
+    runs-on: [self-hosted, fast]
     env:
       BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
 
diff --git a/.github/workflows/ui-self-hosted.yml b/.github/workflows/ui-self-hosted.yml
new file mode 100644
index 000000000..64a4519c6
--- /dev/null
+++ b/.github/workflows/ui-self-hosted.yml
@@ -0,0 +1,118 @@
+name: UI (self-hosted)
+
+# these are the same as ui.yml, but with self-hosted runners
+# the runners come with pre-installed Playwright browsers version: 1.56.1
+# the jobs are much lighter because they don't need to install node and playwright browsers
+
+on:
+  workflow_dispatch:
+    inputs:
+      sha:
+        description: 'Commit SHA1 to build'
+        required: false
+        type: string
+  push:
+    branches:
+      - master
+    paths: [
+      '.github/workflows/ui-self-hosted.yml',
+      '.github/workflows/ui-build.yml',
+      'tools/ui/**.*',
+      'tools/server/tests/**.*'
+    ]
+  pull_request:
+    types: [opened, synchronize, reopened]
+    paths: [
+      '.github/workflows/ui-self-hosted.yml',
+      '.github/workflows/ui-build.yml',
+      'tools/ui/**.*',
+      'tools/server/tests/**.*'
+    ]
+
+env:
+  LLAMA_LOG_COLORS: 1
+  LLAMA_LOG_PREFIX: 1
+  LLAMA_LOG_TIMESTAMPS: 1
+  LLAMA_LOG_VERBOSITY: 10
+
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}-${{ github.head_ref || github.run_id }}
+  cancel-in-progress: true
+
+jobs:
+  ui-build:
+    name: Build static output
+    uses: ./.github/workflows/ui-build.yml
+
+  ui-checks:
+    name: Checks
+    needs: ui-build
+    runs-on: [self-hosted, PLAYWRIGHT]
+    continue-on-error: true
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v6
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
+
+      - name: Install dependencies
+        id: setup
+        run: npm ci
+        working-directory: tools/ui
+
+      - name: Run type checking
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run check
+        working-directory: tools/ui
+
+      - name: Run linting
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run lint
+        working-directory: tools/ui
+
+      - name: Run Client tests
+        if: ${{ always() }}
+        run: npm run test:client
+        working-directory: tools/ui
+
+      - name: Run Unit tests
+        if: ${{ always() }}
+        run: npm run test:unit
+        working-directory: tools/ui
+
+  e2e-tests:
+    name: E2E Tests
+    needs: ui-build
+    runs-on: [self-hosted, PLAYWRIGHT]
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v6
+        with:
+          fetch-depth: 0
+          ref: ${{ github.event.inputs.sha || github.event.pull_request.head.sha || github.sha || github.head_ref || github.ref_name }}
+
+      - name: Install dependencies
+        id: setup
+        run: npm ci
+        working-directory: tools/ui
+
+      - name: Build application
+        if: ${{ always() && steps.setup.conclusion == 'success' }}
+        run: npm run build
+        working-directory: tools/ui
+
+      - name: Build Storybook
+        if: ${{ always() }}
+        run: npm run build-storybook
+        working-directory: tools/ui
+
+      - name: Run UI tests
+        if: ${{ always() }}
+        run: npm run test:ui -- --testTimeout=60000
+        working-directory: tools/ui
+
+      - name: Run E2E tests
+        if: ${{ always() }}
+        run: npm run test:e2e
+        working-directory: tools/ui
diff --git a/.github/workflows/ui-ci.yml b/.github/workflows/ui.yml
similarity index 95%
rename from .github/workflows/ui-ci.yml
rename to .github/workflows/ui.yml
index 761a93194..b5e745718 100644
--- a/.github/workflows/ui-ci.yml
+++ b/.github/workflows/ui.yml
@@ -1,4 +1,4 @@
-name: CI (UI)
+name: UI
 
 on:
   workflow_dispatch:
@@ -11,14 +11,16 @@ on:
     branches:
       - master
     paths: [
-      '.github/workflows/ui-ci.yml',
+      '.github/workflows/ui.yml',
+      '.github/workflows/ui-build.yml',
       'tools/ui/**.*',
       'tools/server/tests/**.*'
     ]
   pull_request:
     types: [opened, synchronize, reopened]
     paths: [
-      '.github/workflows/ui-ci.yml',
+      '.github/workflows/ui.yml',
+      '.github/workflows/ui-build.yml',
       'tools/ui/**.*',
       'tools/server/tests/**.*'
     ]
@@ -39,7 +41,7 @@ jobs:
     uses: ./.github/workflows/ui-build.yml
 
   ui-checks:
-    name: UI Checks
+    name: Checks
     needs: ui-build
     runs-on: ubuntu-latest
     continue-on-error: true
diff --git a/.github/workflows/update-ops-docs.yml b/.github/workflows/update-ops-docs.yml
index 2ab06eb98..6e8bc1aa0 100644
--- a/.github/workflows/update-ops-docs.yml
+++ b/.github/workflows/update-ops-docs.yml
@@ -3,18 +3,20 @@ name: Update Operations Documentation
 on:
     push:
         paths:
+            - '.github/workflows/update-ops-docs.yml'
             - 'docs/ops.md'
             - 'docs/ops/**'
             - 'scripts/create_ops_docs.py'
     pull_request:
         paths:
+            - '.github/workflows/update-ops-docs.yml'
             - 'docs/ops.md'
             - 'docs/ops/**'
             - 'scripts/create_ops_docs.py'
 
 jobs:
     update-ops-docs:
-        runs-on: ubuntu-slim
+        runs-on: [self-hosted, fast, ARM64]
 
         steps:
         - name: Checkout repository