diff --git a/.github/actions/build_cmake/action.yml b/.github/actions/build_cmake/action.yml
index a84089f80..b3bf27e0d 100644
--- a/.github/actions/build_cmake/action.yml
+++ b/.github/actions/build_cmake/action.yml
@@ -58,7 +58,7 @@ runs:
 
         # install base packages for ARM64
         if [ "${{ runner.arch }}" = "ARM64" ]; then
-          conda install -y -q -c conda-forge openblas=0.3.29 gxx_linux-aarch64=14.2 sysroot_linux-aarch64=2.17
+          conda install -y -q -c conda-forge openblas=0.3.33 gxx_linux-aarch64=14.2 sysroot_linux-aarch64=2.17
         fi
 
         # install base packages for X86_64
@@ -77,6 +77,10 @@ runs:
         # and CUDA from cuVS channel for cuVS builds
         elif [ "${{ inputs.cuvs }}" = "ON" ]; then
           conda install -y -q libcuvs=26.02 'cuda-version=12.9' sysroot_linux-64=2.34 -c rapidsai -c rapidsai-nightly -c conda-forge
+          # Clean index cache to prevent sqlite3 "database is locked" errors
+          # from conda-libmamba-solver's shards cache between consecutive installs.
+          # See: https://github.com/conda/conda-libmamba-solver/issues/667
+          conda clean --index-cache 2>/dev/null || true
         fi
 
         # install SVS runtime for SVS builds
diff --git a/conda/faiss-gpu-cuvs/meta.yaml b/conda/faiss-gpu-cuvs/meta.yaml
index 67a70b426..1aa8f2f94 100644
--- a/conda/faiss-gpu-cuvs/meta.yaml
+++ b/conda/faiss-gpu-cuvs/meta.yaml
@@ -61,14 +61,14 @@ outputs:
       host:
         - _openmp_mutex =4.5=2_kmp_llvm  # [x86_64]
         - mkl >=2024.2.2,<2026  # [x86_64]
-        - openblas =0.3.32 # [not x86_64]
+        - openblas =0.3.33 # [not x86_64]
         - libcuvs =26.02
         - cuda-version {{ cuda_constraints }}
         - libsvs-runtime =0.3.0  # [x86_64 and linux]
       run:
         - _openmp_mutex =4.5=2_kmp_llvm  # [x86_64]
         - mkl >=2024.2.2,<2026  # [x86_64]
-        - openblas =0.3.32 # [not x86_64]
+        - openblas =0.3.33 # [not x86_64]
         - cuda-cudart {{ cuda_constraints }}
         - libcublas {{ libcublas_constraints }}
         - libcuvs =26.02
diff --git a/conda/faiss-gpu/meta.yaml b/conda/faiss-gpu/meta.yaml
index 7d29546bf..bd8bb194f 100644
--- a/conda/faiss-gpu/meta.yaml
+++ b/conda/faiss-gpu/meta.yaml
@@ -60,11 +60,11 @@ outputs:
         - cuda-toolkit {{ cudatoolkit }}
       host:
         - mkl >=2024.2.2,<2026  # [x86_64]
-        - openblas =0.3.32 # [not x86_64]
+        - openblas =0.3.33 # [not x86_64]
         - libsvs-runtime =0.3.0  # [x86_64 and linux]
       run:
         - mkl >=2024.2.2,<2026  # [x86_64]
-        - openblas =0.3.32 # [not x86_64]
+        - openblas =0.3.33 # [not x86_64]
         - cuda-cudart {{ cuda_constraints }}
         - libcublas {{ libcublas_constraints }}
         - libsvs-runtime =0.3.0  # [x86_64 and linux]
diff --git a/conda/faiss/meta.yaml b/conda/faiss/meta.yaml
index d87e3378f..91bd78119 100644
--- a/conda/faiss/meta.yaml
+++ b/conda/faiss/meta.yaml
@@ -56,7 +56,7 @@ outputs:
         - mkl-devel >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi
         {% endif %}
-        - libopenblas =0.3.32  # [osx]
+        - libopenblas =0.3.33  # [osx]
       host:
         - python {{ python }}
         - libcxx =20.1.1  # [osx and arm64]
@@ -69,8 +69,8 @@ outputs:
         - mkl >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi =3.12
         {% endif %}
-        - openblas =0.3.32  # [not x86_64]
-        - libopenblas =0.3.32  # [osx]
+        - openblas =0.3.33  # [not x86_64]
+        - libopenblas =0.3.33  # [osx]
       run:
         - python {{ python }}
         - libcxx =20.1.1  # [osx and arm64]
@@ -83,8 +83,8 @@ outputs:
         - mkl >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi =3.12
         {% endif %}
-        - openblas =0.3.32  # [not x86_64]
-        - libopenblas =0.3.32  # [osx]
+        - openblas =0.3.33  # [not x86_64]
+        - libopenblas =0.3.33  # [osx]
     test:
       requires:
         - conda-build =25.1.2
@@ -120,7 +120,7 @@ outputs:
         - mkl >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi =3.12
         {% endif %}
-        - libopenblas =0.3.32  # [osx]
+        - libopenblas =0.3.33  # [osx]
       host:
         - python {{ python }}
         - numpy >=2.0,<3.0
@@ -136,7 +136,7 @@ outputs:
         - mkl >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi =3.12
         {% endif %}
-        - libopenblas =0.3.32  # [osx]
+        - libopenblas =0.3.33  # [osx]
       run:
         - python {{ python }}
         - numpy >=2.0,<3.0
@@ -151,7 +151,7 @@ outputs:
         - mkl >=2024.2.2,<2026  # [x86_64 and win]
         - python_abi =3.12
         {% endif %}
-        - libopenblas =0.3.32  # [osx]
+        - libopenblas =0.3.33  # [osx]
     test:
       requires:
         - numpy >=2.0,<3.0
diff --git a/faiss/svs/IndexSVSVamanaLeanVec.cpp b/faiss/svs/IndexSVSVamanaLeanVec.cpp
index e6b99f9ff..dfcb76e96 100644
--- a/faiss/svs/IndexSVSVamanaLeanVec.cpp
+++ b/faiss/svs/IndexSVSVamanaLeanVec.cpp
@@ -56,7 +56,7 @@ IndexSVSVamanaLeanVec::~IndexSVSVamanaLeanVec() {
         FAISS_ASSERT(status.ok());
         training_data = nullptr;
     }
-    IndexSVSVamana::~IndexSVSVamana();
+    // Base class destructor handles impl cleanup
 }
 
 void IndexSVSVamanaLeanVec::add(idx_t n, const float* x) {