The included changes affect the Camera app, the Dashcam Viewer and Tesla’s blind spot monitoring feature.
Rear Camera
For owners of a refreshed Model Y or original Model Y equipped with AI4, this update brings two improvements to the camera views when reversing.
The backup camera now includes a new toggle button in the top-left corner of the rear camera feed, allowing drivers to switch between the original, narrower view and the newer, wider perspective. When Tesla introduced the wider rear camera in update 2025.14, many users were upset that the view was now further away. Users will now be able to choose their preferred view.
In addition to the new rear camera option, the side camera feeds have also been enlarged to provide a larger field of view.
What’s happening with these changes is that Tesla has uncropped camera views, but they often choose to adjust or crop the camera feed to focus on the most useful areas, making sure that the vehicle’s bumpers and other items don’t occupy unnecessary space. With the changes in this update, Tesla is adjusting the areas of the video that are being cropped.
If you’re interested in what the camera views look like uncropped, you can check out the Camera Previews section in your vehicle that lets you view every raw camera feed.
Another camera improvement is available on update 2025.20.3 for all vehicles that received the updated Dashcam App with AMD Ryzen processors (how to tell if you have Intel or AMD). This includes all Model 3 and Model Y vehicles with AMD Ryzen, as well all 2021+ Model S and Model X vehicles.
When reviewing footage in the Dashcam Viewer app, you can now tap the "expand" button, which looks like two diagonal arrows, to slightly zoom out on the video as it plays. This is a useful feature when trying to get a clearer look at specific details in a Sentry Mode or Dashcam event, such as a license plate, without the hassle of having to export the video file to a computer.
Blind Spot Camera
@max_bracco
Tesla is really focusing on camera improvements with this release. The Blind Spot Camera feature was previously only available whenever the rearview camera was on screen. Turning on the rear camera while your blind spot camera view was enabled would cause the blind spot view to disappear. Likewise, if you turned on your turn signal while the Camera app was open, then the blind spot camera feature would not appear.
With this update, Tesla now allows the Camera app and the blind spot camera feature to be used simultaneously.
Dashcam Viewer Multi-Delete
We recently covered another change to the updated Dashcam App in update 2025.20. This undocumented change allows users to delete multiple videos at once for the first time. This is especially useful if you’re trying to clear up room on your Dashcam USB drive. Previously, you had to delete each video individually, whether you were doing it in the vehicle or through the Tesla app.
Update 2025.20 Rolling Out Now
These camera improvements are rolling out as part of software update 2025.20.3. As with all Tesla software updates, it may take several weeks for this to roll out to your particular vehicle.
Subscribe
Subscribe to our newsletter to stay up to date on the latest Tesla news, upcoming features and software updates.
The Q2 2025 Earnings Call provided us with a lot of information - and much of it related to FSD. There’s a lot of major news to unpack that impacts HW3 owners, AI4 owners, and future purchasers waiting for AI5.
Tesla addressed the previously promised upgrade for HW3 owners, talked about HW4 and also gave us dates of when to expect HW5 (AI5) and beyond.
The HW3 Upgrade
One of the biggest updates from the call was the clarification on the promised hardware upgrade for millions of owners with HW3 (AI3) vehicles. For those hoping for a path to the newer AI4 or future AI5 FSD computer, Tesla has confirmed that this upgrade is on hold until they can solve autonomy.
As we talked about in our HW3 upgrade article, Tesla will not consider offering such a complex and costly upgrade until after FSD Unsupervised is solved and becomes available to customer vehicles. The reasoning is that an upgrade could potentially involve more than just a simple computer swap, potentially requiring changes to other core vehicle components.
The key here is that the power and cooling requirements for HW4 hardware are far greater than those available in HW3 vehicles. If this trend continues with AI5, Tesla will need to build a customized solution for HW3 upgrades that falls somewhere in the middle - capable enough to perform FSD Unsupervised, but still able to fit within the required tolerances.
Camera Upgrades Might Be Needed
In addition, while Tesla has stated that they don’t intend to swap the cameras, this may become a requirement for two reasons. HW3 already has considerable difficulty reading signs, and HW4 is marginally better at doing so. Complex signs, such as “No Right Turn on Red,” could indicate that HW3 vehicles may require a camera upgrade, at least for the primary cameras, if not the side repeaters, bi-pillar, and rear cameras, to comply.
These changes will also require extensive rewiring, as the wiring in HW3 vehicles is not capable of the higher bandwidth required for higher-res cameras. The lack of future-proofing for HW3 vehicles is definitely becoming apparent here.
The second reason is the inclusion of the forward bumper camera on more of Tesla’s fleet. Today, every new vehicle except the Model 3 comes with a bumper camera. While Tesla does not use it for FSD yet, we believe that it will be necessary for low-speed maneuvers in crowded locations and parking lots - the key to making Summon and Banish truly autonomous.
Essentially, for the foreseeable future, HW3 vehicles will continue to be developed on a lagging, separate FSD software branch. Tesla intends to release updates for HW4 vehicles first, then take the time to optimize them, and then release updates for HW3 vehicles once they are ready. Realistically, while HW3 vehicles will still receive FSD updates, it seems that they have reached their physical operational limits and will not achieve the same performance or have all the same features as AI4 or future AI5 vehicles.
HW4’s Unsupervised Capabilities
The earnings call yesterday reinforced the capability of the current-generation HW4 hardware, which is included in all new Tesla vehicles. This is the same hardware that is powering the current Robotaxi FSD fleet that is currently operating and expanding in Austin. This confirms that HW4 is fully capable of supporting a true, driverless experience.
The question going forward will be what the limits of HW4 are, and how soon Tesla is approaching them. Will Tesla provide a smoother path to upgrade HW4 to AI5?
Given the ongoing situation with HW3, we don’t expect that they will. Tesla is contractually bound to provide Unsupervised FSD (autonomy) with the purchase of FSD, but it’s not required to upgrade vehicles so that they can achieve a higher level of safety and comfort. As long as a vehicle is capable of autonomy, then Tesla has met their commitment.
At this point, we don’t expect a clean upgrade path to AI5 - Elon has previously mentioned that it will consume nearly 2-3x the energy of AI4, which means an entirely new electrical and cooling package will be required for AI5, which AI4 vehicles won’t be able to easily support, short of a major retrofit.
However, if you want Unsupervised FSD and are satisfied with not getting a potential upgrade in the future, AI4 is a very compelling choice today. It is already demonstrably capable of Unsupervised FSD, which means you’re getting a vehicle that can and will do autonomy in the future.
If you’re still undecided about a purchase, a 48-month lease could be the most sensible option. Tesla is not intending to introduce AI5 until the end of 2026, and it will likely take months or years for the software divergence between HW4 and AI5 to occur.
AI5 is a Performance Jump; AI6 is a Leap
Tesla also provided the first concrete details on its next-generation FSD hardware, which will be known as AI5. The new computer is projected to be 3 to 5 times more capable than the already powerful HW4 hardware. That’s a massive leap in processing power, which will enable more complex neural networks and faster, more human-like decision-making.
This next-generation hardware isn’t expected to reach mass production until at least the end of 2026 - and the first vehicles receiving it will likely be Tesla’s Cybercab.
Going forward, Tesla intends to use AI6 as a means to closely integrate the training hardware and the vehicle hardware. That means that its Dojo supercomputer chip will live in AI6 vehicles. Two chips in a vehicle or on board an Optimus humanoid robot, while hundreds will live in a training cluster.
Tesla already uses HW4 hardware in Cortex, its Supercompute training cluster at Giga Texas. However, the vast majority of the training hardware is built around Nvidia’s H200 chip, which means Tesla needs to build the training software for the H200, then rebuild the trained models for HW4 hardware.
This change to Dojo training compute and Dojo in-vehicle compute means that Tesla will be able to further optimize the process, simplifying the entire training pipeline.
Closing One Door, Opening Another
The Q2 call was filled with interesting facts that drew solid lines in the sand for what Tesla’s future hardware path will look like. For owners of HW3 vehicles, the path to next-generation hardware is on hold, but at least there’s some clarity.
However, Tesla is offering FSD transfers in many regions worldwide, including North America and Europe, as a way to upgrade customers to newer vehicles equipped with hardware capable of running the latest FSD version.
For current HW4 owners, their vehicles are already equipped with the hardware capable of running in an Unsupervised future, and can look forward to support for at least another few years. In the future, Tesla vehicles will be equipped with even more powerful hardware, including AI5 in late 2026 or early 2027, and AI6 several years later.
If you missed any part of Tesla’s earnings call and its Q&A session, be sure to check out our recap, where we cover everything Tesla talked about, including FSD Unsupervised, the Robotaxi, and much more.
Tesla has released an official LED light strip upgrade for the frunk of the Model 3 and Model Y in the U.S. and Canada. The light strip, priced at $100 USD ($130 CAD), is a great addition to light up one of the most useful storage areas of an EV.
According to Tesla, the new light strip is designed for easy installation by the customer. The light strip connects directly to the vehicle’s power supply using an included adapter, eliminating the need for batteries or wiring harnesses. That makes this a clean OEM setup without the hassle of needing a separate power source.
The light is built to last, and is equipped with a constant current control module and an IP67 dustproof and waterproof rating on both the light itself as well as the adapter, so you don’t need to worry about it getting wet in the rain if you happen to leave your frunk open.
Not a Tesla App
Vehicle Compatibility
The new frunk light strip is available for all Model 3 and Model Y vehicles from model years 2020 to 2024, so it does exclude earlier Model 3s.
For newer vehicles, a revised version of the light strip is being developed, compatible with the updated wiring harness of 2025 and 2026 Model 3 and Model Y vehicles.
The revised version is already available on Tesla’s shop in China, which means its release in North America is likely not too far away.
Sign of More to Come?
This isn’t the first time Tesla has made an OEM version of a popular third-party accessory, such as the custom puddle lights. Another popular third-party accessory is a powered frunk add-on.
Currently, the Cybertruck is the only Tesla vehicle with a powered frunk, and getting an official modification that won’t risk denting your hood would be amazing to see. It would also greatly improve the utility of the frunk, as many people avoid opening it since it has to be manually closed.
This new accessory is a welcome addition to the lineup of accessories for owners looking for a simple and high-quality add-on for their Tesla and and we’re hoping to see more of this in the future.