python 3,14 support

[jdfoxito]

Issue type

Bug

Have you reproduced the bug with TensorFlow Nightly?

Yes

Source

source

TensorFlow version

2.21.0

Custom code

Yes

OS platform and distribution

windows 11

Mobile device

No response

Python version

3.14

Bazel version

No response

GCC/compiler version

No response

CUDA/cuDNN version

No response

GPU model and memory

No response

Current behavior?

C:\nginx>pip install tensorflow
WARNING: Cache entry deserialization failed, entry ignored
ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow

C:\nginx>pip install tf-nightly
ERROR: Could not find a version that satisfies the requirement tf-nightly (from versions: none)
ERROR: No matching distribution found for tf-nightly

C:\nginx> pip install tensorflow-cpu
ERROR: Could not find a version that satisfies the requirement tensorflow-cpu (from versions: none)
ERROR: No matching distribution found for tensorflow-cpu

Standalone code to reproduce the issue

C:\nginx>pip install tensorflow
WARNING: Cache entry deserialization failed, entry ignored
ERROR: Could not find a version that satisfies the requirement tensorflow (from versions: none)
ERROR: No matching distribution found for tensorflow

C:\nginx>pip install tf-nightly
ERROR: Could not find a version that satisfies the requirement tf-nightly (from versions: none)
ERROR: No matching distribution found for tf-nightly

C:\nginx> pip install tensorflow-cpu
ERROR: Could not find a version that satisfies the requirement tensorflow-cpu (from versions: none)
ERROR: No matching distribution found for tensorflow-cpu

Relevant log output

[Kayyuri]

Hi @jdfoxito, Thank you for raising this issue.

This error occurs due to an incompatibility between your TensorFlow and Python versions. To fix this, please install compatible versions. Kindly consult this document for tested build configurations.
Also, Please refer to this relevant issue for further updates.

Thanks!

[jdfoxito]

I’m raising this because our organization is standardizing on Python 3.14 for its enhanced security features and performance gains. It's becoming increasingly difficult to justify staying on older Python versions when the rest of the ecosystem has already updated.
Relying on legacy versions introduces potential security vulnerabilities that we cannot overlook. We would appreciate an update on when we can expect a preview or nightly build for 3.14 to ensure TensorFlow remains a viable part of our high-performance stack.

[jdfoxito]

Python 3.14 has been in the ecosystem for several months now (as a dev/alpha release), and the performance benchmarks—specifically regarding the Faster CPython initiatives and improved memory management—are significant. Our organization aims to leverage these gains to reduce infrastructure costs and improve model inference latency.
Most of our core library stack has already adapted to 3.14. Currently, TensorFlow is the sole blocker preventing us from fully migrating to a more secure and performant environment. Staying on older Python versions is no longer an option due to our internal security compliance and the need for the latest interpreter optimizations.
Could we at least get a nightly build for testing? We are eager to help identify bugs in this new environment."

[mihaimaruseac]

Nightly already supports Python 3.14 (https://pypi.org/project/tf-nightly/#files)

But it seems it's just for linux and macos.

Epic fail.

[jdfoxito]

but i need for windows 11, only 3.12 support i can see, and 3.14 on linux support,