Tesla VP Talks About What's Coming in FSD V15

Close calls on the road unfold in fractions of a second. In one recent case, a driver reported that Full Self-Driving (Supervised) swerved to avoid an oncoming vehicle pulling out blindly from a parking lot, preventing a likely collision before the driver could even tap the horn.

The incident drew a response from Tesla’s VP of AI, Ashok Elluswamy (@aelluswamy), who wrote:

“Glad you are safe. Even earlier prediction of hazards, even faster reaction time and overall significantly better safety and collision avoidance coming as part of the next big upgrade (v15).”

When Tesla released FSD v14.3 in April, the documentation cited 20% faster reactions, and real-world avoidance maneuvers began to look more confident. Earlier this week, Tesla started rolling out FSD v14.3.9, introducing Automatic Collision Evasion that lets the autonomous stack intervene during manual driving to swerve around sudden obstacles. According to Elluswamy, FSD v15 will push these capabilities further.

A 10x Parameter Leap Under the Hood

The shift from v14 to v15 is an architectural upgrade rather than a routine point release. Current builds use roughly one billion parameters, and Elon Musk confirmed a 10x parameter increase with v15, scaling the underlying model to ten billion parameters.

Even today, FSD showcases strong foresight and quick reactions for collision avoidance. A real-world test measured a response to sudden obstacles in just 0.13 seconds. Another comparison reported an average reaction time of 0.45 seconds for FSD 14.3 when evaluated alongside FSD v14 Lite and the older FSD v12.6.4. Owners have also shared numerous examples where FSD avoided crashes, including the one below where it responded to a vehicle breaking a stoplight in 0.2 seconds:

Another video shows an FSD-equipped Tesla reacting to a fast-approaching vehicle from behind, with the driver’s frame-by-frame analysis indicating that FSD began responding about 0.17 seconds before the other vehicle moved toward the Tesla:

Last month, Tesla invited owners to reply with clips of FSD Supervised helping them avoid accidents, resulting in a thread full of videos that emphasize the system’s rapid reaction times:

The larger vision model slated for v15 is expected to make FSD even more capable by processing more information at once and predicting what other drivers may do before they make abrupt moves.

Tesla is already testing FSD v15 in commercial fleets. Robotaxi vehicles are running early builds of FSD v15 across active operating zones, with consumer rollouts expected late this year or early next year. Validating the software is key because Tesla plans to scale its Robotaxi network much faster once FSD v15 arrives.

A Growing Hardware Divide

While the new architecture targets significant safety improvements, it also sets firm hardware boundaries.

Running a ten-billion-parameter neural network requires substantial memory bandwidth and compute. Due to those limits, FSD v15 will be exclusive to vehicles with Hardware 4 (HW4/AI4) computers. Older Hardware 3 vehicles won’t receive the v15 upgrade, and a “v15 Lite” release is not expected.

Instead, HW3 cars will top out on distilled v14 models. Tesla has started rolling out v14.2 Lite to early access testers, so a v14.3 Lite and subsequent v14 Lite iterations may also follow, along with an expansion to Model S and Model X. Ultimately, HW3 owners will need hardware upgrades to access FSD v15 and future unsupervised capabilities.

For HW4 owners, Elluswamy’s comments indicate that the next major software leap aims to make near-misses a non-event well before human reflexes can register, let alone react.

","published":true,"tags":"News","image":{"src":"https://www.notateslaapp.com/img/containers/article_images/fsd-beta/fsd_giftcard.webp/7cfd0633399b2af2d987f0a8fd17d106/fsd_giftcard.jpg","alt":"Tesla VP Talks About What's Coming in FSD V15"}}}