diff --git a/.github/workflows/build-and-release.yaml b/.github/workflows/build-and-release.yaml
index df6201ee7..5cfb6469d 100644
--- a/.github/workflows/build-and-release.yaml
+++ b/.github/workflows/build-and-release.yaml
@@ -82,8 +82,8 @@ jobs:
       - name: Build wheels
         uses: pypa/cibuildwheel@v2.22.0
         env:
-          CIBW_SKIP: "*musllinux* pp*"
-          CIBW_REPAIR_WHEEL_COMMAND: ""
+          CIBW_SKIP: "pp*"
+          CIBW_REPAIR_WHEEL_COMMAND: "LD_LIBRARY_PATH=$PWD/llama_cpp/lib auditwheel repair -w {dest_dir} {wheel}"
           CIBW_ARCHS: "aarch64"
           # Keep native arm64 builds on a portable CPU baseline instead of
           # tuning wheels to the hosted runner.
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 90d79abbf..7c7d20edc 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -7,6 +7,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+- feat: enable arm64 musl builds by @acon96 in #2221
 - feat: Update llama.cpp to ggml-org/llama.cpp@d749821db
 - fix: model fails to load when chat template uses HuggingFace generation tags by @tobocop2 in #2226
 - docs: add contributing guide by @abetlen in #2229