TensorFlow
Ranked 6 of 19 machine learning libraries by recommenders
Early end-user machine learning library, pivotal to fast AI development and widely used. Critics cite hard debugging, breaking changes between versions, and losing ground to PyTorch.
Early end-user machine learning library, pivotal to fast AI development and widely used. Critics cite hard debugging, breaking changes between versions, and losing ground to PyTorch.
56 people gave an opinion
- recommend it
- 20
- criticize it
- 36
- approval share
- 36%
Recommended in 10 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: not recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 20 people recommend itNeeds 10 or more, met.
- 0.5 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 10 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Pivotal to fast AI development
“libraries like TensorFlow which were pivotal to fast development of AI”
Comment from 4 May 2026 in “Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge” “pytorch and tensorflow were also there to facilitate this”
Comment from 22 Dec 2025 in “Ask HN: Why Did Python Win?” Widely used across ML projects
“at least it's one tons of people use”
Comment from 8 Aug 2026 in “Google DeepMind CEO Demis Hassabis is stepping down” “we got tensorflow, pytorch, gpgpus”
Comment from 25 Apr 2026 in “The AI Industry Is Discovering That the Public Hates It” Open-source, scalable deep learning framework
“the lego-like aspect of NN that enabled scalable OSS DL frameworks”
Comment from 25 Apr 2026 in “There Will Be a Scientific Theory of Deep Learning” “Pytorch didn't have to be FOSS, nor Tensorflow.”
Comment from 17 Feb 2026 in “Claude Sonnet 4.6 System Card” Competitive, excels in some use cases
“very capable products, that compete very well in ML space”
Comment from 28 Nov 2025 in “TPUs vs. GPUs and why Google is positioned to win AI race in the long term” “Each of these libraries excels for different use cases.”
Comment from 30 Dec 2025 in “Go Away Python”
What people warn about
Each mark is one comment
Hard to debug static graphs
“The inability to use print debug to tell me the dimensions of my hidden states was 100% why TF was hard”
Comment from 7 Nov 2025 in “Leaving Meta and PyTorch” “Their bet on static compute graphs made writing code verbose and debugging difficult.”
Comment from 7 Nov 2025 in “Leaving Meta and PyTorch” “Tensorflow didn't have that at the time which made debugging practically impossible.”
Comment from 7 Nov 2025 in “Leaving Meta and PyTorch” - Comment from 7 Nov 2025 in “Leaving Meta and PyTorch”
- Comment from 7 Nov 2025 in “Leaving Meta and PyTorch”
Breaking changes across versions
“It's the Tensorflow strategy; deprecate the old version while the new version still doesn't support all the old one's functionality.”
Comment from 19 Feb 2026 in “Gemini 3.1 Pro” “Not being able to load previously trained models with a new version of the library is wildly painful.”
Comment from 7 Nov 2025 in “Leaving Meta and PyTorch” “They killed TF1, upsetting those users, while not having a fully functioning TF2”
Comment from 7 Nov 2025 in “Leaving Meta and PyTorch” Losing ground to PyTorch
“let TensorFlow lose researcher mind share to PyTorch, and caused Google fragmented landscape across TensorFlow and JAX”
Comment from 5 Aug 2026 in “Jeff Dean leaving Alphabet” “Tensorflow is largely dead, it's been years since I've seen a new repo use it.”
Comment from 1 Mar 2026 in “Microgpt” “years later PyTorch ate its lunch, which they eventually admitted with TF 2.0”
Comment from 6 Aug 2026 in “Google DeepMind CEO Demis Hassabis is stepping down”
Other machine learning libraries people recommend
The four most recommended other machine learning libraries, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
See all 19 machine learning libraries, rankedWhat people said
40 of the 63 comments that mention it, newest first. Each links to the original comment.
“there's always something they excel in at some point in time, like TF 1.0”
“obviously went incredibly well”
“which are custom built for those use-cases”
“these other solutions mentioned above are successfully solving for the past ~15y”
“at least it's one tons of people use”
“although I heard everyone say they worked so much better inside Google than outside it, had issues”
“years later PyTorch ate its lunch, which they eventually admitted with TF 2.0”
“let TensorFlow lose researcher mind share to PyTorch, and caused Google fragmented landscape across TensorFlow and JAX”
“Google falling behind the same way it did with Tensorflow, Angular and GCP”
“a lot of ML frameworks (Tensorflow, PyTorch, etc) by default use the GPU relatively suboptimally”
“libraries like TensorFlow which were pivotal to fast development of AI”
“we got tensorflow, pytorch, gpgpus”
“the lego-like aspect of NN that enabled scalable OSS DL frameworks”
“I don't like migrating projects from Tensorflow v.X to Tenserflow v.Y.”
“You'd better hope something like tensorflow isn't one of those dependencies.”
“Tensorflow is largely dead, it's been years since I've seen a new repo use it.”
“It's the Tensorflow strategy; deprecate the old version while the new version still doesn't support all the old one's functionality.”
“Pytorch didn't have to be FOSS, nor Tensorflow.”
“Each of these libraries excels for different use cases.”
“magic error telling that tensorflow version they are installing is not compatible with pytorch version”
“pytorch and tensorflow were also there to facilitate this”
“NumPy (or PyTorch, or TensorFlow, or any Python lib I've encountered) is particularly consistent about this”
“And then lower down we have TensorFlow 0.6.0 release.”
“True, for some definition of first and some definition of modern.”
“TensorFlow was the FIRST modern end-user ML library.”
“Tensorflow is a handy and very useful DSL, but one that is shallow”
“very capable products, that compete very well in ML space”
“exactly the kind of anti-moat everyone has been clamoring for”
“Not being able to load previously trained models with a new version of the library is wildly painful.”
“It made it extremely difficult to debug because you couldn't print tensors easily!”
“The inability to use print debug to tell me the dimensions of my hidden states was 100% why TF was hard”
“there was no way to add debug messages - either your network worked or it didn't”
“required you to first build the network, node by node, then run it”
“why would I use Python to assemble a computational graph when what I really want is to write code and then differentiate through it?”
“They killed TF1, upsetting those users, while not having a fully functioning TF2”
“it won't be able to say the file's name, or that the problem is an input image of the wrong size”
“Tensorflow didn't have that at the time which made debugging practically impossible.”
“Their bet on static compute graphs made writing code verbose and debugging difficult.”
“TF felt like a crude set of bindings to the underlying c++/CUDA workhorse.”
“Python became the new glue for C/CUDA libraries like NumPy, PyTorch and TensorFlow.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.