PyTorch 2.15 RC1 produced for pytorch, torchvision & torchaudio

Release Announcements
1

The core binaries are now available for download and testing on pytorch test channel. Here is the 2.15 release issue tracker for cherry pick submissions.

Reminder of Key Dates
Milestones M1 through M3 are complete.

  • M1: Release Announcement (9/9/26)
  • M2: All PRs landed in PyTorch repo / Feature Submission Closed (2/10/26)
  • M3: Release branch cut, create RC1 for PyTorch, Torchvision and Torchaudio (week of 5/10/26)
  • M4: Release branch finalized, Announce final launch date, Feature classifications published (week of 19/10/26) - Final RC is produced.
  • M4.1: Tutorial drafts submission deadline (23/10/26)
  • M5: External-Facing Content Finalized (27/10/26)
  • M6: Release Day (28/10/26)

Download instructions for testing RC1 for pytorch 2.15.0, torchvision 0.29.2 and torchaudio 2.11.1:

PIP CPU:
Windows/Linux x86 and aarch64/MacOS:

  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/cpu

PIP CUDA 13.2, 13.4
Windows/Linux x86 and aarch64:

  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/cu132
  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/cu134

PIP ROCM 7.14, 10.0

  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/rocm7.14
  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/rocm10.0

PIP XPU
Linux/Windows:

  pip3 install torch==2.15.0 torchvision torchaudio --index-url https://download.pytorch.org/whl/test/xpu

Libtorch
CPU Linux:
https://download.pytorch.org/libtorch/test/cpu/libtorch-shared-with-deps-2.15.0%2Bcpu.zip
CPU Windows (Release version):
https://download.pytorch.org/libtorch/test/cpu/libtorch-win-shared-with-deps-2.15.0%2Bcpu.zip
CPU MacOS:
https://download.pytorch.org/libtorch/test/cpu/libtorch-macos-arm64-2.15.0.zip
CUDA 13.2, 13.4 Linux:
https://download.pytorch.org/libtorch/test/cu132/libtorch-shared-with-deps-2.15.0%2Bcu132.zip
https://download.pytorch.org/libtorch/test/cu134/libtorch-shared-with-deps-2.15.0%2Bcu134.zip
Windows CUDA 13.2, 13.4 (Release version):
https://download.pytorch.org/libtorch/test/cu132/libtorch-win-shared-with-deps-2.15.0%2Bcu132.zip
https://download.pytorch.org/libtorch/test/cu134/libtorch-win-shared-with-deps-2.15.0%2Bcu134.zip
ROCm 7.14, 10.0 Linux:
https://download.pytorch.org/libtorch/test/rocm7.14/libtorch-shared-with-deps-2.15.0%2Brocm7.14.zip
https://download.pytorch.org/libtorch/test/rocm10.0/libtorch-shared-with-deps-2.15.0%2Brocm10.0.zip

Cheers,

Team PyTorch

2

Tested the CPU wheels on Linux x86_64 with Python 3.11 in a clean environment. Basic tensor operations, torch.compile with Inductor, and torchvision transforms ran without issues on 2.15.0 RC1. Will run the broader regression test suite against the CUDA 13.2 test channel binaries next.

1 Like