TensorFlow
Last Updated: July 29, 2026 | By Mihail Sebastian | AI Dictionary
Google's open-source machine learning framework for building, training, and deploying deep learning models, from research code to production and mobile.
What is TensorFlow?
TensorFlow is an open-source machine learning framework from Google Brain, released in 2015, for building, training, and deploying deep learning models.
Its reach is the distinctive part. The same framework covers experimenting in a Python notebook, serving a model behind an API, and running it on a phone or a microcontroller.
How TensorFlow Works
TensorFlow represents a model as a graph of mathematical operations on tensors, the multi-dimensional arrays that hold data and weights. You describe the layers; the framework computes gradients and updates weights during model training, so you never write the calculus yourself.
Most developers work through Keras, TensorFlow’s high-level API, which turns a neural network into a short stack of layer declarations. A trained model then exports to a portable format for serving on servers or running on mobile and edge devices, on CPUs, GPUs, or Google’s TPU chips.
TensorFlow vs PyTorch
The practical difference: TensorFlow leans toward production deployment, while PyTorch leans toward research flexibility. TensorFlow’s original design compiled a model into a fixed graph before running it, which suited deployment; PyTorch builds the graph as the code executes, which suits experimentation.
The two frameworks have borrowed each other’s ideas for years, so the honest answer today is that the choice usually follows the team: what colleagues already know, and which framework the model you are starting from was written in.
| Criterion | TensorFlow | PyTorch |
|---|---|---|
| Origin | Google Brain, open-sourced 2015 | Facebook AI Research (Meta), released 2016 |
| Design emphasis | End-to-end path from training to serving and mobile | Python-native flexibility for experimentation |
| Where it dominates | Production pipelines and edge deployment | Research; most new papers ship PyTorch code |
| High-level interface | Keras | torch.nn plus ecosystem libraries (torchvision, torchaudio) |
Example of TensorFlow
A retailer wants to flag damaged products from warehouse photos. The team defines a convolutional network in Keras, feeds it a labeled training set of damaged and intact items, and lets TensorFlow handle gradient computation and weight updates across a GPU cluster.
After checking accuracy on held-out photos, they export the trained model and deploy it to the scanners on the warehouse floor. That end-to-end path, from notebook experiment to a model running on site hardware, is TensorFlow’s home ground.
Related AI terms: PyTorch · Framework · Deep Learning · Tensors · TPU
Did you like the TensorFlow gist?
Learn about 250+ need-to-know artificial intelligence terms in the AI Dictionary.
Mihail Sebastian — Writes about AI governance, regulation, and the technology behind them. Placeholder bio — replace with a real credential line. About