diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index 443ac1d..3197a2a 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -1,9 +1,8 @@
 on:
   workflow_dispatch:
   push:
-    # Sequence of patterns matched against refs/tags
-    # tags:
-    #   - 'v*' # Push events to matching v*, i.e. v1.0, v20.15.10
+    tags:
+      - 'v*' # Push events to matching v*, i.e. v1.0, v20.15.10
 
 name: Creating release
 
@@ -14,37 +13,8 @@ env:
   CIBW_ARCHS_MACOS: "x86_64 universal2"
 
 jobs:
-  build_wheels:
-    name: Build wheels on ${{ matrix.os }}
-    runs-on: ${{ matrix.os }}
-    strategy:
-      fail-fast: false
-      matrix:
-        os: [ubuntu-20.04, windows-2019, macos-10.15]
-
-    steps:
-      - uses: actions/checkout@v2
-
-      - uses: actions/setup-python@v2
-        name: Setting up Python
-        with:
-          python-version: '3.8'
-
-      - name: Set up QEMU
-        if: runner.os == 'Linux'
-        uses: docker/setup-qemu-action@v1
-        with:
-          platforms: all
-
-      - name: Build & test wheels
-        uses: pypa/cibuildwheel@v2.2.2
-        env:
-          CIBW_ARCHS_LINUX: auto aarch64 ppc64le
-
-      - uses: actions/upload-artifact@v2
-        with:
-          path: ./wheelhouse/*.whl
-
+  # Build & test simple source release before wasting hours building and
+  # testing the binary build matrix.
   sdist:
     name: Creating source release
     runs-on: ubuntu-latest
@@ -74,6 +44,42 @@ jobs:
         with:
           path: dist/*.tar.gz
 
+  build_wheels:
+    needs: [sdist]
+    name: "[${{ strategy.job-index }}/${{ strategy.job-total }}] py${{ matrix.py }} on ${{ matrix.os }}"
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: true
+      matrix:
+        os: [ubuntu-20.04, windows-2019, macos-10.15]
+        # cp - CPython
+        # pp - PyPy
+        py: ["cp36", "cp37", "cp38", "cp39", "cp310", "pp37"]
+
+    steps:
+      - uses: actions/checkout@v2
+
+      - uses: actions/setup-python@v2
+        name: Setting up Python
+        with:
+          python-version: '3.8'
+
+      - name: Set up QEMU
+        if: runner.os == 'Linux'
+        uses: docker/setup-qemu-action@v1
+        with:
+          platforms: all
+
+      - name: Build & test wheels
+        uses: pypa/cibuildwheel@v2.2.2
+        env:
+          CIBW_ARCHS_LINUX: auto aarch64 ppc64le
+          CIBW_BUILD: "${{ matrix.py }}-*"
+
+      - uses: actions/upload-artifact@v2
+        with:
+          path: ./wheelhouse/*.whl
+
   upload_all:
     needs: [build_wheels, sdist]
     name: Uploading built packages to pypi for release.
@@ -84,10 +90,10 @@ jobs:
         name: artifact
         path: dist
 
-    # - uses: pypa/gh-action-pypi-publish@v1.4.2
-    #   with:
-    #     user: ${{ secrets.PYPI_USERNAME }}
-    #     password: ${{ secrets.PYPI_PASSWORD }}
+    - uses: pypa/gh-action-pypi-publish@v1.4.2
+      with:
+        user: ${{ secrets.PYPI_USERNAME }}
+        password: ${{ secrets.PYPI_PASSWORD }}
 
   build_documentation:
     name: Building & uploading documentation.
@@ -114,6 +120,6 @@ jobs:
         run: |
           cd docs && make clean && make html
 
-      # - name: Publishing documentation
-      #   run: |
-      #     ghp-import -f -n -c pysimdjson.tkte.ch -p docs/_build/html
+      - name: Publishing documentation
+        run: |
+          ghp-import -f -n -c pysimdjson.tkte.ch -p docs/_build/html
diff --git a/CHANGELOG.md b/CHANGELOG.md
new file mode 100644
index 0000000..7f4a8ad
--- /dev/null
+++ b/CHANGELOG.md
@@ -0,0 +1,10 @@
+# Changelog
+
+## 4.0.3
+
+- Updated upstream simdjson library to 1.0.2.
+- Binary wheels are now tested and available for py3.6-py3.10 on x86[_64],
+  ppc64le and aarch64.
+  - Under Linux, wheels are built for both manylinux and musl (Alpine).
+  - Under Linux, wheels are also available for PyPy3.7.
+- Minor documentation fixes.
diff --git a/docs/index.rst b/docs/index.rst
index 0718aa7..0cb29b3 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -25,7 +25,7 @@ If SIMD instructions are unavailable a fallback parser is used, making
 pysimdjson safe to use anywhere.
 
 Bindings are currently tested on OS X, Linux, and Windows for Python version
-3.6 to 3.9.
+3.6 to 3.10.
 
 Installation
 ------------
@@ -39,25 +39,28 @@ with no further requirements:
 
 Binary wheels are available for the following:
 
-+-------------+-------+-------+-------+-------+
-|             |     x86_64            |  ARM  |
-+-------------+-------+-------+-------+-------+
-| Interpreter | OS X  |  Win  | Linux | Linux |
-+=============+=======+=======+=======+=======+
-| CPython 3.6 | Yes   | Yes   | Yes   | Yes   |
-+-------------+-------+-------+-------+-------+
-| CPython 3.7 | Yes   | Yes   | Yes   | Yes   |
-+-------------+-------+-------+-------+-------+
-| CPython 3.8 | Yes   | Yes   | Yes   | Yes   |
-+-------------+-------+-------+-------+-------+
-| CPython 3.9 | Yes   | Yes   | Yes   | Yes   |
-+-------------+-------+-------+-------+-------+
++--------------+-------+-------+-------+---------+---------+
+|              |     x86_64            |  ARM64  | PowerPC |
++--------------+-------+-------+-------+---------+---------+
+| Interpreter  | OS X  |  Win  | Linux | Linux   | Linux   |
++==============+=======+=======+=======+=========+=========+
+| CPython 3.6  | Yes   | Yes   | Yes   | Yes     | Yes     |
++--------------+-------+-------+-------+---------+---------+
+| CPython 3.7  | Yes   | Yes   | Yes   | Yes     | Yes     |
++--------------+-------+-------+-------+---------+---------+
+| CPython 3.8  | Yes   | Yes   | Yes   | Yes     | Yes     |
++--------------+-------+-------+-------+---------+---------+
+| CPython 3.9  | Yes   | Yes   | Yes   | Yes     | Yes     |
++--------------+-------+-------+-------+---------+---------+
+| CPython 3.10 | Yes   | Yes   | Yes   | Yes     | Yes     |
++--------------+-------+-------+-------+---------+---------+
 
 When binary wheels are not available, a C++11 (or better) compiler is required
-when installing.
+when installing in order to build the underlying simdjson library.
 
 If you would prefer to always install pysimdjson from source even when
-pre-compiled binaries are available, use:
+pre-compiled binaries are available (to take advantage of a newer compiler for
+example), use:
 
 .. code::
 
diff --git a/docs/performance.rst b/docs/performance.rst
index a9fee8d..2d12b9c 100644
--- a/docs/performance.rst
+++ b/docs/performance.rst
@@ -67,6 +67,7 @@ to re-use the parser.
 
     for i in range(0, 100):
         doc = parser.parse(b'{"a": "b"}')
+        del doc
 
 This will drastically reduce the number of allocations being made, as it will
 reuse the existing buffer when possible. If it's too small, it'll grow to fit.
diff --git a/setup.py b/setup.py
index 14bb57e..5d4466a 100644
--- a/setup.py
+++ b/setup.py
@@ -78,7 +78,7 @@
 setup(
     name='pysimdjson',
     packages=find_packages(),
-    version='4.0.2',
+    version='4.0.3',
     description='simdjson bindings for python',
     long_description=long_description,
     long_description_content_type='text/markdown',
@@ -94,7 +94,7 @@
         'Development Status :: 3 - Alpha',
         'Intended Audience :: Developers',
     ],
-    python_requires='>3.4',
+    python_requires='>3.5',
     extras_require={
         # Dependencies for package release.
         'release': [