Keras Reviews & Product Details

Keras is an open-source software library that provides a Python interface for artificial neural networks. It serves as an interface for TensorFlow, streamlining the building and training of deep-learning models through high-level APIs designed for fast experimentation.

What company sizes and industries is Keras the best fit for?

Keras appears to suit positive users across small, mid-sized, and enterprise organizations, with use spanning software, technology, research, and education. The strongest fit signals cluster in these segments:

Small businesses: Common among users building and experimenting with neural-network models.
Mid-market organizations: A recurring fit for model development and research work.
Enterprise organizations: Used for deep-learning projects, including computer vision and other applied modeling.
Software and IT services: Prominent fields for development and machine-learning work.
Research and higher education: A notable setting for experimentation, coursework, and academic projects.

What customer support options does Keras provide, and how do users find the overall support experience?

Keras support is often experienced through its documentation and community, though getting help with technical issues can be less straightforward. Users value the learning resources and community guidance, while debugging backend problems, interpreting errors, or resolving specialized issues can take extra effort.

What tools and platforms does Keras integrate with, and how do these integrations help users get more value from the product?

Keras integrates with TensorFlow, and the integration helps simplify model-building workflows. It offers 1 integration: TensorFlow. Users find that working through Keras can reduce low-level coding and make it quicker to prototype and develop neural-network models.

What features or capabilities could make Keras even more useful?

Keras could be more useful with stronger customization, clearer troubleshooting, and more consistent performance across demanding workflows. Recurring improvement areas include:

More model flexibility: Expand options for custom layers, architectures, metrics, and other advanced model changes without requiring lower-level code.
Clearer debugging: Improve error messages, logs, and guidance so configuration and backend issues are easier to diagnose.
Performance and compatibility: Make training performance and compatibility across backends, versions, and environments more dependable.

G2 reviews are authentic and verified.

Here's how.

Verified User in Renewables & Environment

Small-Business (50 or fewer emp.)

Originally submitted 4/7/2026

"A Versatile Tool for Machine Learning, Computer Vision, and Deep Learning"

4/5

What do you like best about Keras?

It’s a versatile tool for machine learning, computer vision, and deep learning. Review collected by and hosted on G2.com.

What do you dislike about Keras?

It has compatibility issues, and on some machines it can be difficult to install and set up. Review collected by and hosted on G2.com.

Aakash Kumar A.

Data Scientist

Enterprise (> 1000 emp.)

9/13/2023

"Best DL Framework"

5/5

What do you like best about Keras?

keras is one of the prominent deep learning framework, it is easy to implement and provides great a amount of important functionalities which helps developer to achieve maximum accuracy Review collected by and hosted on G2.com.

What do you dislike about Keras?

There is nothing to dislike in keras except few things like it still haven't upgraded with the latest functionalities such as nlp and generative AI which are some important tools nowadays Review collected by and hosted on G2.com.

Subham S.

Data Scientist

Mid-Market (51-1000 emp.)

8/18/2022

"User-friendly and effective high-level neural-network API"

5/5

What do you like best about Keras?

There are a lot of reasons to like Keras:

1. This open-source deep-learning library is designed to provide fast experimentation with deep neural networks.

2. Keras provides the flexibility to run on top of CNTK, TensorFlow, and Theano.

3. It is focused on being modular, user-friendly, readable, and extensible.

4. Keras provides the power to build deep neural networks using fewer lines of code, and this amazes me the most.

5. Since Keras was adopted and integrated into TensorFlow in mid-2017, we can leverage its power by deploying trained models to production thanks to the TensorFlow Serving framework.

6. Keras has excellent access to reusable code and tutorials, which makes it extremely suitable even for beginners.

7. Since Keras runs on top of TensorFlow, it can be equipped with single or multiple GPUs for faster computations. Review collected by and hosted on G2.com.

What do you dislike about Keras?

There are a few reasons for disliking Keras:

1. Keras is not very customizable on its own. While researching different algorithms or working on multi-dimensional matrices, we still need scikit-learn, OpenCV, or Tensorflow for performing such operations.

2. Sometimes the errors are difficult to debug since finding error logs is difficult.

For these reasons alone, Keras is still one of the most popular and favorite libraries for statisticians, data scientists, ML engineers, etc. Review collected by and hosted on G2.com.

What problems is Keras solving and how is that benefiting you?

As a data scientist, I have to very frequently build different state-of-the-art deep learning models, and Keras is one such library without which this task seems impossible. One of my recent projects involved Time-Series Forecasting, which involved building different state-of-the-art Deep Learning models like Transformers, LSTMs, etc., which was done using Keras and Tensorflow.

If anyone wants to go through a simple implementation using Keras: https://github.com/SubhamIO/KERAS-on-MNIST-dataset Review collected by and hosted on G2.com.

Bassel M.

ADAS Machine Learning Engineer

Mid-Market (51-1000 emp.)

Originally submitted 1/31/2021

"Keras has always provided with all the tools for machine learning"

5/5

What do you like best about Keras?

I like the simplicity of building neural networks Review collected by and hosted on G2.com.

What do you dislike about Keras?

The difficulty of implementing my customized metric Review collected by and hosted on G2.com.

Sathesh R.

Managed Services Integration Analyst

Enterprise (> 1000 emp.)

9/15/2022

"Deep Learning made easy and a wonderful library that does offer a lot!"

4/5

What do you like best about Keras?

Best wrapper API available out there for Neural networks. You need not have to be an expert programmer, it does offer what you need to get the job done and it is an open source. Integrates well with tensor flow. Its native to python and I come up with python background, it does make my coding world a lot easier. Implementing a neural network would take hours of coding, but Keras has made it simpler with few lines of code and it is easily understandable. Review collected by and hosted on G2.com.

What do you dislike about Keras?

Understanding of the log traces to fix a problem takes time, as you would have to understand the way it is traced and written, that would take time because of limited documentation. As a python developer I find it to be easier to use, but it doesnt provide other language support, it could be a problem for a long term development. It doesnt offer a great backend support as it is limited. Review collected by and hosted on G2.com.

Paweł W.

Software Engineer @ Creators

Small-Business (50 or fewer emp.)

10/9/2022

"Open source tool for managing artificial neural networks"

4/5

What do you like best about Keras?

First of all Keras is a complete API for managing neural networks and is an open source tool. I find its API extremely convenient to use - definitely simpler to use than PyTorch Review collected by and hosted on G2.com.

What do you dislike about Keras?

It might get slow for some complicated use cases, so if you are aiming for speed and efficiency then probably PyTorch would be a better choice. Review collected by and hosted on G2.com.

Dipak K.

Senior Research Fellow (PhD)

Small-Business (50 or fewer emp.)

Originally submitted 6/8/2022

"User friendly high level framework for developing deep neural networks"

4.5/5

What do you like best about Keras?

easy and quick implementation of a variety of neural network models. simple and easy to learn with vast support from the Keras community and documentation. I like the most about Keras is the high-level framework and runs on top of TensorFlow with single or multiple GPUs for faster computations. availability of pre-trained models such as VGGNET, RESNET, etc. Review collected by and hosted on G2.com.

What do you dislike about Keras?

Preprocessing of the signals or images is still not widely used due to lack of customization. one needs to use additional tools such as Scikit-learn to do the proper preprocessing. Issues in the low-level backend can not be targeted and finding those error logs is difficult. Other than these issues Keras is widely famous in the AI field. Review collected by and hosted on G2.com.

Ashish T.

System Engineer

Mid-Market (51-1000 emp.)

9/27/2022

"An excellent neural network library to run our models"

4/5

What do you like best about Keras?

It is easier to use and setup on most of the backend systems like tensor flow and pytorch. This provides a lot of operational freedom to developers to experiment. Review collected by and hosted on G2.com.

What do you dislike about Keras?

Some external integration is difficult to implement on the system and requires assistance from consultants. Initial setup on windows OS is also a bit challenging. Review collected by and hosted on G2.com.

Gourav S.

Open Source Contributor

Small-Business (50 or fewer emp.)

9/13/2022

"Great for beginners and for somewhat advanced as well."

4/5

What do you like best about Keras?

Keras is amazing with its documentation and I've used it on Google collab. It worked super well, models were meeting expectations. Review collected by and hosted on G2.com.

What do you dislike about Keras?

Might not be that great as compared to alternatives, when it comes to speed, it's somewhat slow. Review collected by and hosted on G2.com.