EVwire brief: Tesla has begun rolling out FSD (Supervised) v14.3, with users reporting noticeably faster reactions and smoother driving behavior.
Tesla said the update delivers a 20% improvement in reaction time, alongside broader upgrades to its neural network training and perception systems.
Before we get to the compilation from the Tesla owners out there, here are the full release notes for FSD (Supervised) v14.3:
Full Self-Driving (Supervised) v14.3 includes
Upgraded the Reinforcement Learning (RL) stage of training the FSD neural network, resulting in improvements in a wide variety of driving scenarios.
Upgraded the neural network vision encoder, improving understanding in rare and low-visibility scenarios, strengthening 3D geometry understanding, and expanding traffic sign understanding.
Rewrote the AI compiler and runtime from the ground up with MLIR, resulting in 20% faster reaction time and improving model iteration speed.
Mitigated unnecessary lane biasing and minor tailgating behaviors.
Increased decisiveness of parking spot selection and maneuvering.
Improved parking location pin prediction, now shown on a map with a (P) icon.
Enhanced response to emergency vehicles, school buses, right-of-way violators, and other rare vehicles.
Improved handling of small animals by focusing RL training on harder examples and adding rewards for better proactive safety.
Improved traffic light handling at complex intersections with compound lights, curved roads, and yellow light stopping – driven by training on hard RL examples sourced from the Tesla fleet.
Improved handling for rare and unusual objects extending, hanging, or leaning into the vehicle path by sourcing infrequent events from the fleet.
Improved handling of temporary system degradations by maintaining control and automatically recovering without driver intervention, reducing unnecessary disengagements.
Upcoming Improvements
Expand reasoning to all behaviors beyond destination handling.
Add pothole avoidance.
Improve driver monitoring system sensitivity with better eye gaze tracking, eye wear handling, and higher accuracy in variable lighting conditions.
Early user feedback highlights FSD (Supervised) v14.3’s more responsive behavior
Initial reactions from users have been largely positive, with many pointing to improved responsiveness.
Longtime FSD tester AIDRIVR, after real-world tests of the system, described FSD v14.3’s reaction time as downright inhuman:
He also mentioned that Tesla’s “20% improvement” in reaction time is an understatement:
This improved reaction time has been observed even for people who suddenly dart into the road:
Or, as observed by longtime Tesla owner @BLKMLD3, reacting to cars that suddenly back into you in the parking lot:
Some also observed that Mad Max mode has become more refined. It’s still fast and assertive, but it feels like a more mature driver now:
More importantly, FSD v14.3 reacts very well to being cut off by other drivers, as shown in this video from Dirty Tesla:
Our friend Sawyer Merritt also praised FSD v14.3’s parking capabilities, as it now backs and pulls forward more quickly and confidently:
Source: Tesla AI on X
DIG DEEPER into the Tesla industry news with our dedicated TESLAWIRE page. And don’t forget to subscribe to our EV industry newsletter to join 14,000+ EV geeks.
