Overview
TensorFlow 2.21, released on March 6, 2026, drops Python 3.9 and removes the TensorBoard dependency.
Main Features
Python 3.9 dropped
TensorFlow 2.21 requires Python 3.10 or later, enabling the use of modern language features in the internal codebase.
python
import tensorflow as tf
model = tf.keras.Sequential([
tf.keras.layers.Dense(128, activation='relu'),
tf.keras.layers.Dense(10, activation='softmax'),
])
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy')
print(f'TF version: {tf.__version__}')TensorBoard dependency removed
TensorBoard is no longer automatically installed with TensorFlow. It must be installed separately via pip install tensorboard.