Full Self Driving

FSD Version History › FSD 11.4

Tesla FSD 11.4

Released

Overview

FSD 11.4 is a major single-stack release merging the highway and city driving stacks into one unified neural network pipeline. This unified model handles all driving scenarios without mode-switching.

What's new in FSD 11.4

  • Unified highway and city driving stack — one neural network for all roads
  • Elimination of the mode-switching boundary between highway and city logic
  • Improved consistency as vehicles transition between road types
  • Better overall calibration of speed and position on unfamiliar roads
  • Foundation for the V12 end-to-end neural architecture

Known issues

  • Early reports of some highway-to-city transition edge cases
  • Some reduction in performance in certain previously well-handled scenarios during the merge

Community feedback

FSD 11.x single-stack is viewed as the precursor to V12. Merging the stacks eliminates a class of bugs tied to mode transitions and sets the stage for the full end-to-end V12 redesign.

How does Tesla FSD 11.4 compare?

See how Tesla FSD stacks up against Waymo, Mobileye, and other autonomous driving systems.

View FSD vs Waymo comparison →

Frequently asked questions

What is new in FSD 11.4?

Unified highway and city driving stack — one neural network for all roads. Elimination of the mode-switching boundary between highway and city logic. Improved consistency as vehicles transition between road types. Better overall calibration of speed and position on unfamiliar roads. Foundation for the V12 end-to-end neural architecture

What are the known issues in FSD 11.4?

Early reports of some highway-to-city transition edge cases. Some reduction in performance in certain previously well-handled scenarios during the merge

What do FSD 11.4 users say?

FSD 11.x single-stack is viewed as the precursor to V12. Merging the stacks eliminates a class of bugs tied to mode transitions and sets the stage for the full end-to-end V12 redesign.

Related autonomous driving research

Explore the academic papers behind FSD's end-to-end neural network approach: