diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index f109294c..45304662 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -15,17 +15,42 @@ jobs:
       fail-fast: true
       matrix:
         os: [ubuntu-latest]
-        python-version: ["3.9", "3.10", "3.12"]
+        # NOTE(stes): We test against the oldest still supported version of python (3.10),
+        # the python version of the latest Ubuntu LTS release (3.12)
+        # and the latest python version CEBRA supports fully (3.13).
+        #
+        # Python version chart:
+        #   https://devguide.python.org/versions/
+        # Python Ubuntu LTS:
+        #   https://documentation.ubuntu.com/ubuntu-for-developers/reference/availability/python/
+        python-version: ["3.10", "3.12", "3.13"]
         # We aim to support the versions on pytorch.org
         # as well as selected previous versions on
         # https://pytorch.org/get-started/previous-versions/
-        torch-version: ["2.4.0", "2.6.0"]
+        torch-version: ["2.6.0", "2.9.1"]
         sklearn-version: ["latest"]
         include:
+          # windows test with standard config
           - os: windows-latest
-            torch-version: 2.4.0
-            python-version: "3.10"
+            torch-version: 2.6.0
+            python-version: "3.12"
             sklearn-version: "latest"
+
+          # legacy sklearn (several API differences)
+          - os: ubuntu-latest
+            torch-version: 2.6.0
+            python-version: "3.12"
+            sklearn-version: "legacy"
+
+          # TODO(stes): latest torch and python
+          # requires a PyTables release compatible with
+          # python 3.14. Update when new version is released.
+          #- os: ubuntu-latest
+          #  torch-version: 2.9.1
+          #  python-version: "3.14"
+          #  sklearn-version: "latest"
+
+          # legacy support
           - os: ubuntu-latest
             torch-version: 2.4.0
             python-version: "3.10"
@@ -82,7 +107,7 @@ jobs:
           make check_for_binary
 
       - name: Run pytest tests
-        timeout-minutes: 10
+        timeout-minutes: 15
         run: |
           make test
 
diff --git a/.github/workflows/latest.yml b/.github/workflows/latest.yml
new file mode 100644
index 00000000..a549d717
--- /dev/null
+++ b/.github/workflows/latest.yml
@@ -0,0 +1,48 @@
+name: Python package (latest)
+
+# This is a variant of the workflow in build.yml to check incompatibilities for the
+# very latest versions of Python and the core dependencies.
+
+on:
+  push:
+    branches:
+      - main
+  pull_request:
+    branches:
+      - main
+
+jobs:
+  build:
+    timeout-minutes: 30
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Cache dependencies
+        id: pip-cache
+        uses: actions/cache@v4
+        with:
+          path: ~/.cache/pip
+          key: pip-os_latest
+
+      - name: Checkout code
+        uses: actions/checkout@v2
+
+      - name: Set up Python 3.14
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.14"
+
+      - name: Install package
+        run: |
+          python -m pip install --upgrade pip setuptools wheel --pre
+          python -m pip install --pre torch --extra-index-url https://download.pytorch.org/whl/cpu
+          pip install --pre '.[dev]'
+
+      - name: Run pytest tests
+        timeout-minutes: 15
+        run: |
+          make test_reduced
+
+      - name: Build package
+        run: |
+          make build
diff --git a/Makefile b/Makefile
index f5696026..44270e68 100644
--- a/Makefile
+++ b/Makefile
@@ -39,6 +39,21 @@ test_all: clean_test
 test_fast: clean_test
 	python -m pytest --ff --ignore cebra/grid_search.py -m "not requires_dataset" tests cebra --runfast
 
+# Reduced variant of the test suite. These tests should pass even if only the core package without any
+# addtional dataset/integration dependencies are installed.
+test_reduced: clean_test
+	python -m pytest --ff --ignore cebra/grid_search.py \
+		--ignore tests/test_attribution.py \
+		--ignore tests/test_dlc.py \
+		--ignore tests/test_grid_search.py \
+		--ignore tests/test_integration_xcebra.py \
+		--ignore tests/test_load.py \
+		--ignore tests/test_plotly.py \
+		--ignore cebra/attribution \
+		--ignore cebra/integrations \
+		-m "not requires_dataset" tests cebra --runfast
+
+
 # Run failed test firsts, using a single worker (for debugging)
 test_debug: clean_test
 	python -m pytest -vvv -x --ff -m "not requires_dataset" tests
diff --git a/cebra/integrations/matplotlib.py b/cebra/integrations/matplotlib.py
index 844844fd..a322acb7 100644
--- a/cebra/integrations/matplotlib.py
+++ b/cebra/integrations/matplotlib.py
@@ -68,6 +68,11 @@ def __init__(self, axis: Optional[matplotlib.axes.Axes], figsize: tuple,
         if axis is None:
             self.fig = plt.figure(figsize=figsize, dpi=dpi)
 
+    def __del__(self):
+        if hasattr(self, "fig") and self.fig is not None:
+            plt.close(self.fig)
+            self.fig = None
+
     @abc.abstractmethod
     def _define_ax(
             self, axis: Optional[matplotlib.axes.Axes]) -> matplotlib.axes.Axes:
diff --git a/cebra/integrations/sklearn/cebra.py b/cebra/integrations/sklearn/cebra.py
index 0f056b45..abead081 100644
--- a/cebra/integrations/sklearn/cebra.py
+++ b/cebra/integrations/sklearn/cebra.py
@@ -21,6 +21,7 @@
 #
 """Define the CEBRA model."""
 
+import importlib.metadata
 import itertools
 from typing import (Callable, Dict, Iterable, List, Literal, Optional, Tuple,
                     Union)
@@ -28,7 +29,6 @@
 import numpy as np
 import numpy.typing as npt
 import packaging.version
-import importlib.metadata
 import sklearn
 import sklearn.utils.validation as sklearn_utils_validation
 import torch
@@ -46,9 +46,12 @@
 
 # NOTE(stes): From torch 2.6 onwards, we need to specify the following list
 # when loading CEBRA models to allow weights_only = True.
+# NOTE(stes): "numpy.dtypes.Int32DType" was added due to this issue with
+# windows (https://github.com/AdaptiveMotorControlLab/CEBRA/pull/281#issuecomment-3764185072)
+# on build (windows-latest, torch 2.6.0, python 3.12, latest sklearn)
 CEBRA_LOAD_SAFE_GLOBALS = [
     cebra.data.Offset, torch.torch_version.TorchVersion, np.dtype,
-    np.dtypes.Float64DType, np.dtypes.Int64DType
+    np.dtypes.Int32DType, np.dtypes.Float64DType, np.dtypes.Int64DType
 ]
 
 
@@ -1398,7 +1401,7 @@ def save(self,
                                 np.__version__,
                             'sklearn_version':
                                 importlib.metadata.distribution("scikit-learn"
-                                                              ).version
+                                                               ).version
                         }
                     }, filename)
             else:
diff --git a/setup.cfg b/setup.cfg
index 0bfa5f84..dad9ac26 100644
--- a/setup.cfg
+++ b/setup.cfg
@@ -28,11 +28,10 @@ packages = find:
 where =
     - .
     - tests
-python_requires = >=3.9
+python_requires = >=3.10
 install_requires =
     joblib
     numpy<2.0;platform_system=="Windows"
-    numpy<2.0;platform_system!="Windows" and python_version<"3.10"
     numpy;platform_system!="Windows" and python_version>="3.10"
     literate-dataclasses
     scikit-learn
@@ -57,6 +56,7 @@ datasets =
     # additional data loading dependencies
     hdf5storage # for creating .mat files in new format
     openpyxl # for excel file format loading
+    tables # for hdf5 file format loading
 integrations =
     pandas
     plotly
@@ -104,8 +104,8 @@ dev =
     pytest-benchmark
     pytest-xdist
     pytest-timeout
-    pytest-sphinx
-    tables
+    # NOTE(stes): https://github.com/twmr/pytest-sphinx/issues/69
+    pytest-sphinx @ https://github.com/twmr/pytest-sphinx/archive/refs/tags/v0.7.0.tar.gz
     licenseheaders
     interrogate
     # TODO(stes) Add back once upstream issue
diff --git a/tests/_util.py b/tests/_util.py
index 42dd54cb..96c3f0d8 100644
--- a/tests/_util.py
+++ b/tests/_util.py
@@ -20,7 +20,9 @@
 # limitations under the License.
 #
 import collections.abc as collections_abc
+import functools
 
+import packaging.version
 import pytest
 import sklearn.utils.estimator_checks
 import torch
@@ -63,17 +65,20 @@ def parametrize_slow(arg_names, fast_arguments, slow_arguments):
 
 
 def parametrize_with_checks_slow(fast_arguments, slow_arguments):
-    fast_params = [
-        list(
-            sklearn.utils.estimator_checks.check_estimator(
-                fast_arg, generate_only=True))[0] for fast_arg in fast_arguments
-    ]
-    slow_params = [
-        list(
-            sklearn.utils.estimator_checks.check_estimator(
-                slow_arg, generate_only=True))[0] for slow_arg in slow_arguments
-    ]
-    return parametrize_slow("estimator,check", fast_params, slow_params)
+
+    # NOTE(stes): See https://github.com/AdaptiveMotorControlLab/CEBRA/issues/280, sklearn API changed in 1.6.
+    if packaging.version.parse(
+            sklearn.__version__) <= packaging.version.parse("1.6"):
+        generate_checks = functools.partial(
+            sklearn.utils.estimator_checks.check_estimator, generate_only=True)
+    else:
+        generate_checks = sklearn.utils.estimator_checks.estimator_checks_generator
+
+    def _generate_params(args):
+        return [next(generate_checks(arg)) for arg in args]
+
+    return parametrize_slow("estimator,check", _generate_params(fast_arguments),
+                            _generate_params(slow_arguments))
 
 
 def parametrize_device(func):
diff --git a/tests/_utils_deprecated.py b/tests/_utils_deprecated.py
index 5c533f26..a5a2aaaf 100644
--- a/tests/_utils_deprecated.py
+++ b/tests/_utils_deprecated.py
@@ -38,12 +38,6 @@ def cebra_transform_deprecated(cebra_model,
         >>> embedding = cebra_model.transform(dataset)
 
     """
-    warnings.warn(
-        "The method is deprecated "
-        "but kept for testing puroposes."
-        "We recommend using `transform` instead.",
-        DeprecationWarning,
-        stacklevel=2)
 
     sklearn_utils_validation.check_is_fitted(cebra_model, "n_features_")