@stefanolson@SawyerMerritt@hanspjoh@TeslaAUNZ Oh you brought it outright, I understand now. I paid for last year beginning September and waited till end of September and that wait was not fun. Every morning checking the car.
@RobGrieves They are definitely improving it. I know it very well with 25000k on Gold Coast roads, its planning is improving and so its assertiveness. Still get in wrong lane sometimes. cant wait for 14 mad max or even hurry
@MicrosoftHelps Hi, I’m in Australia and need help with a Microsoft 365 subscription billing issue. The account uses an expired custom-domain email, so I can’t receive the verification code, but I’m still being charged. What is the correct support path to cancel or recover the subscription? I have been going around in circles for 6 months.
FSD saw the pedestrian I completely missed 👀
Sitting at the lights Palm Beach. Light goes green, I start moving… idiot walks straight out from behind the Bus on the left. I didn’t see him. FSD did — instant, calm, perfect reaction.
I trust it every day. Down Under needs V14
#FSD #Tesla #FullSelfDriving #FSDEveryday #TeslaAustralia
🚨 Big Tesla FSD update just landed in Australia.
My Model 3 is downloading 2026.8.3.10 now — Tesla describes it as a major update to FSD Supervised v13.2.9.
I filmed the download and first impressions. Next step: real-world Gold Coast testing.
Is this the update that moves FSD Australia forward?
Video coming soon. 🇦🇺⚡
@Josh143574359 youre right. That is crazy. I know the current version very well. Driven over 20,000ks with it around the Gold Coast. Will know by tomorrow f anything has changed in FSD itself.
@Josh143574359 The release notes say FSD (Supervised) v13 Down Under upgrades every part of the end-to-end driving network. but then provides the old release notes
Official Release Notes
FSD (Supervised) v13.2.9 Down Under FSD (Supervised) v13 Down Under upgrades every part of the end-to-end driving network. Includes:
36 Hz, full-resolution AI4 video inputs
Native AI4 inputs and neural network architectures
4.2x data scaling
5x training compute scaling (enabled by the Cortex cluster)
Reduced photon-to-control latency by 2x
Integrated unpark, reverse, and park capabilities
Improved reward predictions for collision avoidance
Improved camera cleaning
Redesigned controller for smoother, more accurate tracking
Dynamic routing around road closures, which displays them along an affected route when they are detected by the fleet
This is what Grok says about 2026.8.3.10
Elaboration on Each Major Part (Engineer’s Perspective)
FSD (Supervised) v13.2.9 Down Under – The Big One for Your HW4 Model 3 This is a major neural network refresh for Australia/NZ on HW4 cars. Even though the visible FSD version stays 13.2.9, the underlying model is significantly upgraded.
36 Hz, full-resolution AI4 video inputs: HW4 cameras now feed the neural net at 36 frames per second (vs lower rates before). More data per second = smoother, more responsive decisions.
Native AI4 inputs and neural network architectures: The model was rebuilt from the ground up for HW4’s higher-res cameras and processing power (no more “downgrading” from HW3 designs).
4.2x data scaling + 5x training compute (Cortex cluster): Tesla trained this version with vastly more real-world Australian driving data and compute. This is why you’re seeing it more assertive, faster acceleration, and more confident — the neural net has seen far more examples.
Reduced photon-to-control latency by 2x: From light hitting the camera to steering/throttle output — twice as fast. Feels more “alive” and less laggy.
Integrated unpark/reverse/park: Parking manoeuvres now use the same end-to-end network instead of separate code → more natural.
Improved reward predictions for collision avoidance: The AI is better at scoring “how bad would a crash be here?” — leads to smarter gap-taking and avoidance.
Redesigned controller: The final output layer that turns AI decisions into steering/throttle/brake commands is smoother → less twitchy.
Dynamic routing around road closures: Fleet data now shows temporary closures on your map in real time.
Grok describe it as such see below.
Elaboration on Each Major Part (Engineer’s Perspective)
FSD (Supervised) v13.2.9 Down Under – The Big One for Your HW4 Model 3This is a major neural network refresh for Australia/NZ on HW4 cars. Even though the visible FSD version stays 13.2.9, the underlying model is significantly upgraded.
36 Hz, full-resolution AI4 video inputs: HW4 cameras now feed the neural net at 36 frames per second (vs lower rates before). More data per second = smoother, more responsive decisions.
Native AI4 inputs and neural network architectures: The model was rebuilt from the ground up for HW4’s higher-res cameras and processing power (no more “downgrading” from HW3 designs).
4.2x data scaling + 5x training compute (Cortex cluster): Tesla trained this version with vastly more real-world Australian driving data and compute. This is why you’re seeing it more assertive, faster acceleration, and more confident — the neural net has seen far more examples.
Reduced photon-to-control latency by 2x: From light hitting the camera to steering/throttle output — twice as fast. Feels more “alive” and less laggy.
Integrated unpark/reverse/park: Parking manoeuvres now use the same end-to-end network instead of separate code → more natural.
Improved reward predictions for collision avoidance: The AI is better at scoring “how bad would a crash be here?” — leads to smarter gap-taking and avoidance.
Redesigned controller: The final output layer that turns AI decisions into steering/throttle/brake commands is smoother → less twitchy.
Dynamic routing around road closures: Fleet data now shows temporary closures on your map in real time.
Upc