You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Let me answer directly: Most likely not, on the contrary, your performance may decrease.
PyTorch 2.7.1, which you are currently using, is considered one of the most stable and fastest versions, especially for video processing (ComfyUI, WanVideo, etc.). Although PyTorch 2.9 (and 2.9.1) is new (October/November 2025), many users are experiencing serious problems in your field.
We can make a comparison according to your profile and interests (Video AI/ComfyUI) as follows:
Why should you stay with 2.7.1? (Recommended)
Speed Difference: Tests show that version 2.7.1 can work about 2 times faster than 2.9 on video models (especially in the VAE decode stage). For example, some users reported that the same job took 300 seconds in 2.7 and 600 seconds in 2.9.
"Eager Mode" Performance: ComfyUI still mostly uses "Eager mode", and 2.7 is more responsive in this mode. 2.9 is more optimized for "torch.compile" mode, which may not work stably in every workflow.
Compatibility: Many of the "custom nodes" you use may not be fully optimized for 2.9 and CUDA 13 yet.
Conclusion: If your system is currently running smoothly, do not upgrade at all. PyTorch 2.7.1 is currently considered the "sweet spot" for video AI work. Moving to 2.9 may slow down your system.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
Gemini:
Let me answer directly: Most likely not, on the contrary, your performance may decrease.
PyTorch 2.7.1, which you are currently using, is considered one of the most stable and fastest versions, especially for video processing (ComfyUI, WanVideo, etc.). Although PyTorch 2.9 (and 2.9.1) is new (October/November 2025), many users are experiencing serious problems in your field.
We can make a comparison according to your profile and interests (Video AI/ComfyUI) as follows:
Why should you stay with 2.7.1? (Recommended)
Speed Difference: Tests show that version 2.7.1 can work about 2 times faster than 2.9 on video models (especially in the VAE decode stage). For example, some users reported that the same job took 300 seconds in 2.7 and 600 seconds in 2.9.
"Eager Mode" Performance: ComfyUI still mostly uses "Eager mode", and 2.7 is more responsive in this mode. 2.9 is more optimized for "torch.compile" mode, which may not work stably in every workflow.
Compatibility: Many of the "custom nodes" you use may not be fully optimized for 2.9 and CUDA 13 yet.
Conclusion: If your system is currently running smoothly, do not upgrade at all. PyTorch 2.7.1 is currently considered the "sweet spot" for video AI work. Moving to 2.9 may slow down your system.
Beta Was this translation helpful? Give feedback.
All reactions