[diffusion] model: support minimax-h3 (#33275)

Co-authored-by: zhenaozhenfu <zhenaozhenfu@minimaxi.com>
Co-authored-by: BBuf <1182563586@qq.com>
Co-authored-by: andyluo7 <andy.luo@amd.com>
Co-authored-by: Zijie Xia <zijie_xia@icloud.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: chao-xue <877184285@qq.com>
Co-authored-by: Cursor <cursoragent@cursor.com>

.github/workflows/pr-test-multimodal-gen.yml

@@ -435,6 +435,55 @@ jobs:
       - uses: ./.github/actions/upload-cuda-coredumps
         if: failure()
 
+  multimodal-gen-test-4-h100:
+    # Temporarily disabled while the 4-gpu-h100 runner is unstable
+    if: ${{ false }}
+    runs-on: 4-gpu-h100
+    timeout-minutes: 90
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          ref: ${{ inputs.git_ref || github.sha }}
+
+      - uses: ./.github/actions/check-pr-test-health
+
+      - uses: ./.github/actions/check-maintenance
+
+      - name: Download artifacts
+        if: inputs.sgl_kernel == 'true'
+        uses: actions/download-artifact@v4
+        with:
+          path: python/sglang/kernels/aot/dist/
+          merge-multiple: true
+          pattern: wheel-python3.10-cuda*
+
+      - name: Install dependencies
+        timeout-minutes: 20
+        run: |
+          CUSTOM_BUILD_SGL_KERNEL=${{inputs.sgl_kernel}} bash scripts/ci/cuda/ci_install_dependency.sh diffusion
+
+      - name: Run MiniMax-H3 PR smoke test
+        timeout-minutes: 45
+        env:
+          RUNAI_STREAMER_MEMORY_LIMIT: 0
+          SGLANG_TEST_WAIT_SECS: 1800
+          SGLANG_DIFFUSION_ARTIFACT_DIR: ${{ github.workspace }}/diffusion-failures
+        run: |
+          cd python
+          python3 sglang/multimodal_gen/test/run_suite.py --suite 4-gpu-h100
+
+      - name: Upload diffusion failure artifacts
+        if: always()
+        uses: actions/upload-artifact@v4
+        with:
+          name: diffusion-failures-${{ github.job }}-${{ github.run_attempt }}
+          path: diffusion-failures/
+          if-no-files-found: ignore
+
+      - uses: ./.github/actions/upload-cuda-coredumps
+        if: failure()
+
   multimodal-gen-unit-test:
     if: |
       ((github.event_name == 'schedule' || inputs.test_parallel_dispatch == 'true') || (inputs.caller_needs_failure != 'true' && !cancelled())) &&