---
id: tf-models-nightly
version: "2.21.0.dev20260217"
license: Apache 2.0
license_treatment: permissive
maintenance: active
---

# tf-models-nightly — TensorFlow Official Models

License: permissive · Maintenance: active · Downloads: 107.3K/mo

## What it is and what it does

tf-models-nightly is a nightly build distribution of the TensorFlow Model Garden, a repository of reference implementations for state-of-the-art machine learning models. It bundles official TensorFlow models maintained and kept current with TensorFlow 2 APIs, research implementations, and the Orbit training library for custom training loops. The package includes 28 runtime dependencies spanning computer vision tools (opencv-python-headless, Pillow), NLP utilities (sentencepiece, seqeval), data handling (pandas, numpy, scipy), and specialized TensorFlow components (tensorflow-text-nightly, tf-hub-nightly, tensorflow-model-optimization).

Because this is a nightly build, it reflects the latest changes from the master branch daily, making it suitable for developers who want cutting-edge model implementations and are willing to accept potential API instability. The package is designed to demonstrate best practices for modeling in TensorFlow 2. Installation is straightforward via pip, though users should be aware that nightly dependencies may introduce breaking changes.

Use it for:

- Prototyping and experimenting with state-of-the-art model architectures for research projects
- Building NLP pipelines using TensorFlow text models and training utilities from the repository
- Learning TensorFlow best practices by studying reference implementations of SOTA models
- Implementing custom training loops with Orbit for distributed training across CPU, GPU, and TPU
- Accessing pre-built computer vision models for image classification, detection, or segmentation tasks

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides state-of-the-art TensorFlow model implementations and training solutions across computer vision, NLP, and other domains with official and research implementations.

Yes, with conditions. Install if you need access to cutting-edge TensorFlow model implementations and are comfortable with nightly build instability. The active maintenance, permissive license, and low install friction make it valuable for research and experimentation. Avoid if you require API stability or production-grade guarantees.

## Install

pip install tf-models-nightly
uv add tf-models-nightly
poetry add tf-models-nightly

## Installing tf-models-nightly

Before you install: Low install friction with a pure-Python wheel distribution. Active maintenance with recent commits and 77654 repository stars indicate ongoing development, though the nightly build nature means API stability is not guaranteed.

License in practice: Apache 2.0 permissive license allows commercial and private use with minimal restrictions, making it suitable for both research and production applications.

Quickstart:

pip3 install tf-models-nightly

import os
os.environ['PYTHONPATH'] += ":/path/to/models"

from tensorflow_models import models

Requires Python >=3.7 and TensorFlow nightly builds (tf-nightly, tf-keras-nightly, tensorflow-text-nightly, tf-hub-nightly) which are development versions and may have breaking changes.

Verify before relying:

- Specific model performance benchmarks or accuracy metrics for the included implementations
- Which SOTA models are currently included and their training status on TensorBoard.dev
- Compatibility guarantees between nightly TensorFlow dependencies and this package's release cycle
- API stability and backward compatibility policy for nightly releases

## Package facts

- License: Apache 2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 107.3K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

tensorflow model implementations, sota deep learning models, tensorflow training examples, computer vision tensorflow, nlp tensorflow models, tensorflow model garden, tensorflow best practices, tensorflow-models, nightly-build, ml-reference-implementations

[View on SkillFed](https://skillfed.io/packages/tf-models-nightly) · [View on PyPI](https://pypi.org/project/tf-models-nightly/)