diff --git a/.github/CODEOWNERS b/.github/CODEOWNERS
index 9bac2756b8b..19188206ea0 100644
--- a/.github/CODEOWNERS
+++ b/.github/CODEOWNERS
@@ -9,7 +9,7 @@ LICENSE @NVIDIA/modelopt-setup-codeowners
 LICENSE_HEADER @NVIDIA/modelopt-setup-codeowners
 pyproject.toml @NVIDIA/modelopt-setup-codeowners
 SECURITY.md @NVIDIA/modelopt-setup-codeowners
-tox.ini @NVIDIA/modelopt-setup-codeowners
+noxfile.py @NVIDIA/modelopt-setup-codeowners
 uv.lock @NVIDIA/modelopt-setup-codeowners
 
 # Library
diff --git a/.github/workflows/code_quality.yml b/.github/workflows/code_quality.yml
index 723f3af1209..cd877d18d24 100644
--- a/.github/workflows/code_quality.yml
+++ b/.github/workflows/code_quality.yml
@@ -24,4 +24,4 @@ jobs:
         with:
           extra_args: --results=verified,unknown
       - name: Run code quality checks
-        run: pip install tox && tox -e pre-commit-all
+        run: pip install nox uv && nox -s pre_commit_all
diff --git a/.github/workflows/gpu_tests.yml b/.github/workflows/gpu_tests.yml
index d2b06229888..95d633ec7c7 100644
--- a/.github/workflows/gpu_tests.yml
+++ b/.github/workflows/gpu_tests.yml
@@ -41,16 +41,16 @@ jobs:
           files: |
             .github/workflows/gpu_tests.yml
             modelopt/**
+            noxfile.py
+            pyproject.toml
             tests/gpu/**
             tests/gpu_megatron/**
-            tests/gpu_regression/**
             tests/gpu_trtllm/**
+            tests/gpu_regression/**
             examples/speculative_decoding/**
             examples/dataset/**
             modelopt_recipes/general/speculative_decoding/**
             tools/launcher/**
-            pyproject.toml
-            tox.ini
           fail_on_initial_diff_error: true
   wait-checks:
     needs: [check-file-changes]
@@ -101,14 +101,8 @@ jobs:
           COVERAGE_PROCESS_START: ${{ github.workspace }}/pyproject.toml
           COVERAGE_FILE: ${{ github.workspace }}/.coverage
         run: |
-          # nemo containers use uv venvs which is not compatible with tox-current-env, so run tests directly
-          if [[ "${{ matrix.example }}" == "gpu_megatron" ]]; then
-            python -m pip install -e .[hf,dev-test]
-            python -m pytest tests/gpu_megatron --cov
-          else
-            python -m pip install tox tox-current-env
-            COV_ARGS="--cov" python -m tox -e cuda13-${{ matrix.example }} --current-env
-          fi
+          pip install nox
+          nox -s ${{ matrix.example }}
       - name: Upload GPU coverage to Codecov
         uses: codecov/codecov-action@v5
         with:
diff --git a/.github/workflows/pages.yml b/.github/workflows/pages.yml
index 6789142f97f..eda19778594 100644
--- a/.github/workflows/pages.yml
+++ b/.github/workflows/pages.yml
@@ -27,7 +27,7 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Build docs
-        run: pip install tox && tox -e build-docs
+        run: pip install nox uv && nox -s docs
       - name: Upload docs artifact
         if: github.event_name == 'push' || github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
         uses: actions/upload-artifact@v4
@@ -47,7 +47,7 @@ jobs:
       - uses: ./.github/actions/ubuntu-setup
       - name: Build docs
         if: github.event.action != 'closed'
-        run: pip install tox && tox -e build-docs
+        run: pip install nox uv && nox -s docs
       - name: Deploy / remove PR preview
         uses: rossjrw/pr-preview-action@v1
         with:
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index e3e62d78def..35edb9e53ef 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -13,13 +13,11 @@ jobs:
     steps:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
-      - name: Install dependencies
-        run: pip install tox
       - name: Run basic unit tests
-        run: tox -e py312-torch29-tf_latest-unit
+        run: pip install nox uv && nox -s "unit-3.12(torch_211, tf_latest)"
       - name: Build Wheel
         run: |
-          tox -e build-wheel
+          nox -s build_wheel
           echo "WHEEL_PATH=$(find dist -name "*.whl" | head -n 1)" >> $GITHUB_ENV
       - name: Upload GitHub Release Artifact
         env:
diff --git a/.github/workflows/unit_tests.yml b/.github/workflows/unit_tests.yml
index afa57fc3a86..29bd48ebd07 100644
--- a/.github/workflows/unit_tests.yml
+++ b/.github/workflows/unit_tests.yml
@@ -9,9 +9,9 @@ on:
     paths:
       - ".github/workflows/unit_tests.yml"
       - "modelopt/**"
-      - "tests/unit/**"
+      - "noxfile.py"
       - "pyproject.toml"
-      - "tox.ini"
+      - "tests/unit/**"
       - "tools/launcher/**"
   schedule:
     - cron: "0 0 * * *" # Nightly
@@ -38,7 +38,10 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Run unit tests
-        run: pip install tox && COV_ARGS="--cov" tox -e py312-torch211-tf_latest-unit
+        env:
+          COVERAGE_PROCESS_START: ${{ github.workspace }}/pyproject.toml
+          COVERAGE_FILE: ${{ github.workspace }}/.coverage
+        run: pip install nox uv && nox -s "unit-3.12(torch_211, tf_latest)"
       - name: Upload coverage reports to Codecov
         uses: codecov/codecov-action@v5
         with:
@@ -57,8 +60,7 @@ jobs:
         with:
           python-version: "3.12"
       - name: Run unit tests (without coverage)
-        # Some issues with torch 2.10 on Windows, so using 2.9 for now
-        run: pip install tox && tox -e py312-torch29-tf_latest-unit
+        run: pip install nox uv && nox -s "unit-3.12(torch_211, tf_latest)"
   multi-py:
     if: github.event_name == 'pull_request'
     needs: [linux]
@@ -74,7 +76,7 @@ jobs:
         with:
           python-version: "3.${{ matrix.py }}"
       - name: Run unit tests
-        run: pip install tox && tox -e py3${{ matrix.py }}-torch211-tf_latest-unit
+        run: pip install nox uv && nox -s "unit-3.${{ matrix.py }}(torch_211, tf_latest)"
   multi-torch:
     if: github.event_name == 'pull_request'
     needs: [linux]
@@ -88,7 +90,7 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Run unit tests
-        run: pip install tox && tox -e py312-torch${{ matrix.torch }}-tf_latest-unit
+        run: pip install nox uv && nox -s "unit-3.12(torch_${{ matrix.torch }}, tf_latest)"
   multi-transformers:
     if: github.event_name == 'pull_request'
     needs: [linux]
@@ -102,7 +104,7 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Run unit tests
-        run: pip install tox && tox -e py312-torch211-tf_${{ matrix.tf }}-unit
+        run: pip install nox uv && nox -s "unit-3.12(torch_211, tf_min)"
   launcher:
     if: github.event_name == 'pull_request'
     needs: [linux]
@@ -133,7 +135,7 @@ jobs:
       - uses: actions/checkout@v6
       - uses: ./.github/actions/ubuntu-setup
       - name: Run unit tests
-        run: pip install tox && tox -e py312-partial-unit-${{ matrix.test-env }}
+        run: pip install nox uv && nox -s "partial_unit(subset='${{ matrix.test-env }}')"
   unit-pr-required-check:
     # Run even if some jobs are skipped
     if: ${{ github.event_name == 'pull_request' && always() }}
diff --git a/.vscode/settings.json b/.vscode/settings.json
index 0e8465ad38d..5a1c278b0f7 100644
--- a/.vscode/settings.json
+++ b/.vscode/settings.json
@@ -19,7 +19,6 @@
         ".mypy_cache": true,
         ".pytest_cache": true,
         ".ruff_cache": true,
-        ".tox": true,
         "**/__pycache__/**": true,
         "**/*.pyc": true,
         "**/runs": true,
diff --git a/CLAUDE.md b/CLAUDE.md
index d07775f3a85..4af38586788 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -37,9 +37,9 @@ Primarily Python codebase with optional C++/CUDA extensions supporting PyTorch,
 | Pattern match | `pytest tests/unit -k "test_quantize"` |
 | Lint + format (all files) | `pre-commit run --all-files` |
 | Lint (diff only) | `pre-commit run --from-ref origin/main --to-ref HEAD` |
-| Run via tox (CPU unit) | `tox -e py312-torch210-tf_latest-unit` |
-| Build docs | `tox -e build-docs` |
-| Build wheel | `tox -e build-wheel` |
+| Run via nox (CPU unit) | `nox -s "unit-3.12(torch_211, tf_latest)"` |
+| Build docs | `nox -s docs` |
+| Build wheel | `nox -s build_wheel` |
 
 ## Architecture
 
@@ -104,7 +104,7 @@ A **recipe** is a declarative YAML specification of an optimization configuratio
 | `modelopt_recipes/general/ptq/` | Built-in PTQ recipe YAML files |
 | `pyproject.toml` | Optional dependency groups (`[onnx]`, `[hf]`, `[all]`, `[dev]`); ruff, mypy, pytest, bandit, and coverage config |
 | `.pre-commit-config.yaml` | Pre-commit hooks (ruff, mypy, clang-format, license headers) |
-| `tox.ini` | Test environment definitions |
+| `noxfile.py` | Test session definitions |
 
 ## Design Patterns
 
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 23f07850a6b..53e879a7bf3 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -89,7 +89,7 @@ We use [pytest](https://docs.pytest.org/) for all tests. For any new features /
 - `tests/gpu_trtllm`: Fast GPU-based unit tests for the core ModelOpt library for TensorRT-LLM features. In most cases, they should not take more than a few seconds to run.
 - `tests/examples`: Integration tests for ModelOpt examples. They should not take more than a few minutes to run. Please refer to [example test README](./tests/examples/README.md) for more details.
 
-Please refer to [tox.ini](./tox.ini) for more details on how to run the tests and their dependencies.
+Please refer to [noxfile.py](./noxfile.py) for more details on how to run the tests and their dependencies.
 
 ## ✍️ Signing your work
 
diff --git a/noxfile.py b/noxfile.py
new file mode 100644
index 00000000000..3d6d9d66eca
--- /dev/null
+++ b/noxfile.py
@@ -0,0 +1,193 @@
+# SPDX-FileCopyrightText: Copyright (c) 2024 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+"""Nox session definitions for testing, linting, docs, and wheel builds.
+
+Usage:
+    python -m pip install nox uv                                                    # install nox and uv (once)
+    nox -l                                                                          # list all sessions
+    nox -s gpu_megatron                                                             # run a GPU session (inside container)
+    nox -s "unit-3.12(torch_211, tf_latest)"                                        # run a specific unit test combination
+    nox -s "unit-3.12(torch_211, tf_latest)" -R                                     # force-recreate venv (e.g. after dep changes)
+    COVERAGE_PROCESS_START=pyproject.toml nox -s "unit-3.12(torch_211, tf_latest)"  # with coverage
+"""
+
+import glob
+import os
+import shutil
+
+import nox
+
+nox.options.default_venv_backend = "uv" if shutil.which("uv") else "virtualenv"
+nox.options.envdir = "/tmp/.nox"
+nox.options.reuse_existing_virtualenvs = True
+
+TORCH_VERSIONS = {
+    "torch_28": "torchvision~=0.23.0",
+    "torch_29": "torchvision~=0.24.0",
+    "torch_210": "torchvision~=0.25.0",
+    "torch_211": "torchvision~=0.26.0",
+}
+
+TRANSFORMERS_VERSIONS = {
+    "tf_latest": None,
+    "tf_min": "transformers~=4.56.0",
+}
+
+
+def _cov_args():
+    """Return --cov when COVERAGE_PROCESS_START is set (CI only)."""
+    return ["--cov"] if os.environ.get("COVERAGE_PROCESS_START") else []
+
+
+# ─── CPU unit tests ───────────────────────────────────────────────────────────
+@nox.session(python=["3.10", "3.11", "3.12", "3.13"])
+@nox.parametrize("tf_ver", [nox.param(k, id=k) for k in TRANSFORMERS_VERSIONS])
+@nox.parametrize("torch_ver", [nox.param(k, id=k) for k in TORCH_VERSIONS])
+def unit(session, torch_ver, tf_ver):
+    """Unit tests — parametrized over torch and transformers versions."""
+    session.install(TORCH_VERSIONS[torch_ver], "-e", ".[all,dev-test]")
+    tf_pin = TRANSFORMERS_VERSIONS[tf_ver]
+    if tf_pin:
+        session.install(tf_pin)
+    session.run("python", "-m", "pytest", "tests/unit", *_cov_args())
+
+
+@nox.session(python="3.12")
+@nox.parametrize("subset", ["onnx", "torch", "torch_deploy"])
+def partial_unit(session, subset):
+    """Unit tests with partial installs."""
+    if subset == "onnx":
+        session.install("torchvision~=0.26.0", ".[onnx,dev-test]")
+        session.run("python", "-m", "pytest", "tests/unit/onnx")
+    elif subset == "torch":
+        session.install("megatron-core", ".[dev-test]")
+        session.run(
+            "python",
+            "-m",
+            "pytest",
+            "tests/unit/torch",
+            "--ignore=tests/unit/torch/deploy",
+            "--ignore=tests/unit/torch/puzzletron",
+        )
+    else:  # torch_deploy
+        session.install(".[onnx,dev-test]")
+        session.run("python", "-m", "pytest", "tests/unit/torch/deploy")
+
+
+# ─── GPU sessions (run inside containers — no new venv) ──────────────────────
+# `venv_backend="none"` skips creating a new venv so the session runs directly in the container's
+# existing Python environment (e.g. /opt/venv in NeMo) instead of an isolated one.
+# Use `python -m pip/pytest` to ensure the container's active venv Python is used,
+# not a stale PATH entry (e.g. NeMo container has pip → /usr/local/bin/pip but python → /opt/venv/bin/python).
+# Container: nvcr.io/nvidia/pytorch:26.01-py3 or later
+@nox.session(venv_backend="none")
+def gpu(session):
+    # tests/gpu/_extensions/test_onnx_extensions.py fails for newer containers
+    # until https://github.com/tbenthompson/cppimport/pull/98
+    session.run(
+        "python",
+        "-m",
+        "pip",
+        "install",
+        "--no-build-isolation",
+        "git+https://github.com/Dao-AILab/fast-hadamard-transform.git",
+    )
+    session.run("python", "-m", "pip", "install", "-e", ".[all,dev-test]")
+    session.run("python", "-m", "pip", "uninstall", "-y", "cupy-cuda12x")
+    session.run("python", "-m", "pip", "install", "cupy-cuda13x")
+    session.run(
+        "python",
+        "-m",
+        "pip",
+        "install",
+        "--no-build-isolation",
+        "git+https://github.com/state-spaces/mamba.git",
+        "git+https://github.com/Dao-AILab/causal-conv1d.git",
+    )
+    session.run("python", "-m", "pytest", "tests/gpu", *_cov_args())
+
+
+# Container: nvcr.io/nvidia/nemo:26.02 or later
+@nox.session(venv_backend="none")
+def gpu_megatron(session):
+    session.run("python", "-m", "pip", "install", "-e", ".[hf,dev-test]")
+    session.run("python", "-m", "pytest", "tests/gpu_megatron", *_cov_args())
+
+
+# Container: nvcr.io/nvidia/pytorch:26.01-py3 or later
+@nox.session(venv_backend="none")
+def gpu_regression(session):
+    session.run("python", "-m", "pip", "install", "-e", ".[hf,dev-test]")
+    session.run("python", "-m", "pytest", "tests/gpu_regression", *_cov_args())
+
+
+# Container: nvcr.io/nvidia/tensorrt-llm/release:1.3.0rc10 or later
+@nox.session(venv_backend="none")
+def gpu_trtllm(session):
+    session.run("python", "-m", "pip", "install", "-e", ".[hf,dev-test]")
+    session.run("python", "-m", "pytest", "tests/gpu_trtllm", *_cov_args())
+
+
+# ─── Code quality ─────────────────────────────────────────────────────────────
+@nox.session
+def pre_commit_all(session):
+    session.install("-e", ".[all,dev-lint]")
+    session.run("pre-commit", "run", "--all-files", "--show-diff-on-failure")
+
+
+@nox.session
+def pre_commit_diff(session):
+    session.install("-e", ".[all,dev-lint]")
+    session.run("pre-commit", "run", "--from-ref", "origin/main", "--to-ref", "HEAD")
+
+
+# ─── Docs ─────────────────────────────────────────────────────────────────────
+@nox.session
+def docs(session):
+    session.install("-e", ".[all,dev-docs]")
+    shutil.rmtree("docs/build", ignore_errors=True)
+    shutil.rmtree("docs/source/reference/generated", ignore_errors=True)
+    with session.chdir("docs"):
+        session.run(
+            "sphinx-build",
+            "source",
+            "build/html",
+            "--fail-on-warning",
+            "--show-traceback",
+            "--keep-going",
+        )
+
+
+@nox.session
+def docs_debug(session):
+    session.install("-e", ".[all,dev-docs]")
+    shutil.rmtree("docs/build", ignore_errors=True)
+    shutil.rmtree("docs/source/reference/generated", ignore_errors=True)
+    with session.chdir("docs"):
+        session.run("sphinx-autobuild", "source", "build/html", "--host", "0.0.0.0")
+
+
+# ─── Wheel build ──────────────────────────────────────────────────────────────
+@nox.session
+def build_wheel(session):
+    shutil.rmtree("build", ignore_errors=True)
+    session.install("twine")
+    session.run("pip", "wheel", "--no-deps", "--wheel-dir=dist", ".")
+    wheels = glob.glob("dist/*.whl")
+    session.run("twine", "check", *wheels)
+    (modelopt_wheel,) = glob.glob("dist/nvidia_modelopt-*.whl")
+    session.install(modelopt_wheel, "-f", "dist")
+    with session.chdir("dist"):
+        session.run("python", "-c", "import modelopt; print(modelopt.__version__)")
diff --git a/pyproject.toml b/pyproject.toml
index 2993759ec10..cf27836f8c9 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -82,7 +82,7 @@ hf = [
     "peft>=0.17.0",
     "sentencepiece>=0.2.1",                                                           # Also implicitly used in test_unified_export_megatron, test_vllm_fakequant_megatron_export
     "tiktoken",
-    "transformers>=4.56",                                                             # Should match modelopt/torch/__init__.py and tox.ini
+    "transformers>=4.56",                                                             # Should match modelopt/torch/__init__.py and noxfile.py
     "wonderwords",
 ]
 
@@ -112,17 +112,17 @@ dev-docs = [
     "sphinx-togglebutton>=0.3.2",
 ]
 dev-test = [
-    "coverage[toml]>=7.13.0",  # a1_coverage.pth for subprocess tracking requires this
+    "coverage[toml]>=7.13.0", # a1_coverage.pth for subprocess tracking requires this
     "pytest",
     "pytest-cov",
     "pytest-instafail",
     "pytest-timeout",
     "timm",
-    "torchprofile==0.0.4",     # optional dependency for modelopt.torch
+    "torchprofile==0.0.4",    # optional dependency for modelopt.torch
     "torchvision",
     "torch-geometric",
-    "tox>4.18",
-    "tox-current-env>=0.0.12",
+    "uv",
+    "nox",
 ]
 # Compound extras via self-references
 all = ["nvidia-modelopt[onnx,hf,puzzletron]"]
@@ -209,6 +209,7 @@ extend-ignore = [
 [tool.ruff.lint.per-file-ignores]
 "__init__.py" = ["F401", "F403"]
 "examples/*" = ["D"]
+"noxfile.py" = ["D", "E501"]
 "tests/*" = ["B017", "D", "E402", "PT012"]
 "*/_[a-zA-Z]*" = ["D"] # Private packages (_abc/*.py) or modules (_xyz.py)
 "*.ipynb" = [
@@ -280,7 +281,7 @@ module = ["examples.*"]
 disable_error_code = ["attr-defined"]
 
 [tool.bandit]
-exclude_dirs = [".github/", "examples/", "tests/"]
+exclude_dirs = [".github/", "examples/", "noxfile.py", "tests/"]
 # Do not change `skips`. It should be consistent with NVIDIA's Wheel-CI-CD bandit.yml config.
 # Use of `# nosec BXXX` requires special approval
 skips = [
diff --git a/tools/launcher/tests/conftest.py b/tools/launcher/tests/conftest.py
index da98e8ad6cb..6d6cb56f635 100644
--- a/tools/launcher/tests/conftest.py
+++ b/tools/launcher/tests/conftest.py
@@ -20,8 +20,8 @@
     uv pip install pytest
     uv run python3 -m pytest tests/ -v
 
-Or via tox from Model-Optimizer root:
-    tox -e py312-launcher
+Or via nox from Model-Optimizer root:
+    nox -s "unit-3.12(torch_211-tf_latest)"
 """
 
 import os
diff --git a/tox.ini b/tox.ini
deleted file mode 100644
index 4f77033b29b..00000000000
--- a/tox.ini
+++ /dev/null
@@ -1,145 +0,0 @@
-[tox]
-envlist=
-    pre-commit-all
-    py312-torch211-tf_latest-unit
-    cuda13-gpu
-    cuda13-gpu_regression
-skipsdist = True
-toxworkdir = /tmp/{env:USER}-modelopt-tox
-passenv =
-    SETUPTOOLS_SCM_PRETEND_VERSION
-
-############################
-# CPU Unit test environments
-############################
-[testenv:{py310,py311,py312,py313}-torch{28,29,210,211}-tf_{min,latest}-unit]
-deps =
-    # torch version auto-selected based on torchvision version
-    torch28: torchvision~=0.23.0
-    torch29: torchvision~=0.24.0
-    torch210: torchvision~=0.25.0
-    torch211: torchvision~=0.26.0
-
-    -e .[all,dev-test]
-
-    # Should match pyproject.toml
-    tf_min: transformers~=4.56.0
-commands =
-    python -m pytest tests/unit {env:COV_ARGS:}
-
-
-#####################################################################
-# Environment to run unit tests with subset of dependencies installed
-#####################################################################
-[testenv:{py310,py311,py312,py313}-partial-unit-{onnx,torch,torch_deploy}]
-allowlist_externals =
-    bash, rm
-deps =
-    # Make sure torch 2.10 is used
-    torchvision~=0.26.0
-
-    # ONNX unit tests heavily rely on torch / torchvision
-    onnx: .[onnx,dev-test]
-    onnx: torchvision
-
-    # Install megatron-core to test torch-only install can still import plugins
-    torch: megatron-core
-    torch: .[dev-test]
-
-    torch_deploy: .[onnx,torch,dev-test]
-commands =
-    onnx: python -m pytest tests/unit/onnx
-    torch: python -m pytest tests/unit/torch --ignore tests/unit/torch/deploy --ignore tests/unit/torch/puzzletron
-    torch_deploy: python -m pytest tests/unit/torch/deploy
-
-
-###########################################################
-# GPU test environments (Should be used with --current-env)
-###########################################################
-# Container: nvcr.io/nvidia/pytorch:26.01-py3 or later
-[testenv:cuda13-gpu]
-commands_pre =
-    # Install deps here so that it gets installed even in --current-env
-    python -m pip install --no-build-isolation git+https://github.com/Dao-AILab/fast-hadamard-transform.git
-    python -m pip install -e .[all,dev-test]
-
-    # Install cupy-cuda13x for INT4 ONNX quantization (default is cupy-cuda12x)
-    python -m pip uninstall -y cupy-cuda12x
-    python -m pip install cupy-cuda13x
-
-    # Install mamba and causal-conv1d for Nemotron tests
-    python -m pip install --no-build-isolation git+https://github.com/state-spaces/mamba.git
-    python -m pip install --no-build-isolation git+https://github.com/Dao-AILab/causal-conv1d.git
-commands =
-    python -m pytest tests/gpu {env:COV_ARGS:}
-
-[testenv:cuda13-gpu_regression]
-commands_pre =
-    python -m pip install -e .[hf,dev-test]
-commands =
-    python -m pytest tests/gpu_regression {env:COV_ARGS:}
-
-# Container: nvcr.io/nvidia/nemo:26.02 or later
-# NOTE: tox is bypassed for this env in CI (see gpu_tests.yml) because tox-current-env is
-# incompatible with uv venvs, and any new tox env would lack NeMo/Megatron packages from /opt/venv.
-# [testenv:cuda13-gpu_megatron]
-# commands_pre =
-#     python -m pip install -e .[hf,dev-test]
-# commands =
-#     python -m pytest tests/gpu_megatron {env:COV_ARGS:}
-
-# Container: nvcr.io/nvidia/tensorrt-llm/release:1.2.0 or later
-[testenv:cuda13-gpu_trtllm]
-commands_pre =
-    # Install deps here so that it gets installed even in --current-env
-    python -m pip install -e .[hf,dev-test]
-commands =
-    python -m pytest tests/gpu_trtllm {env:COV_ARGS:}
-
-#############################################
-# Code quality checks on all files or on diff
-#############################################
-[testenv:{pre-commit}-{all,diff}]
-deps =
-    -e .[all,dev-lint]
-commands =
-    all: pre-commit run --all-files --show-diff-on-failure {posargs}
-    diff: pre-commit run --from-ref origin/main --to-ref HEAD {posargs}
-
-
-#########################
-# Run documentation build
-#########################
-[testenv:{build,debug}-docs]
-allowlist_externals =
-    rm
-deps =
-    -e .[all,dev-docs]
-changedir = docs
-commands_pre =
-    rm -rf build
-    rm -rf source/reference/generated
-commands =
-    sphinx-build source build/html --fail-on-warning --show-traceback --keep-going
-    debug: sphinx-autobuild source build/html --host 0.0.0.0
-
-
-#################
-# Run wheel build
-#################
-[testenv:build-wheel]
-allowlist_externals =
-    bash, cd, rm
-deps =
-    twine
-commands =
-    # Clean build directory to avoid any stale files getting into the wheel
-    rm -rf build
-
-    # Build and check wheel
-    pip wheel --no-deps --wheel-dir=dist .
-    twine check dist/*
-
-    # Install and test the wheel
-    bash -c "find dist -name 'nvidia_modelopt-*.whl' | xargs pip install -f dist"
-    bash -c "cd dist; python -c 'import modelopt; print(modelopt.__version__);'"