The GTA 4 versus GTA 5 driving argument is fifteen years old and has never once been settled with evidence. This is an attempt to settle it with what exists.
Will GTA 6 drive like GTA 4 or GTA 5?
The evidence points toward the GTA 4 end, but nothing confirms it outright. What is confirmed: Rockstar has said its vehicle handling team includes real racing drivers, and Rob Nelson has said the streets in Leonida are narrower and closer to real-world scale, which together with traffic density makes driving take a bit of skill. That is a statement about difficulty, not about a physics model, and the distinction matters. What can be observed rather than inferred is the door deformation in Trailer 1, which is in-engine and behaves like soft deformation rather than the preset damage states GTA 5 used. What cannot be settled is the handling model itself, because a race at Gellhorn International Raceway and a getaway sequence both show cars behaving well under specific conditions, which is exactly what a trailer is for.
What Rockstar has actually confirmed
Two things, both on the record. The vehicle handling team includes real racing drivers, and Rob Nelson has described the streets of Vice City and the wider state of Leonida as narrower and closer to real-world scale, which combined with traffic density makes driving require a bit of skill.
Neither statement describes a physics model. A studio can hire racing drivers to tune arcade handling just as easily as simulation handling, and "takes a bit of skill" is a difficulty claim about navigating tight streets in traffic. Both are meaningful. Neither is the answer people are reading into them.
The door deformation in Trailer 1
This is the strongest observable evidence, and it has been sitting in public since the first trailer. A door deforms in-engine in a way that reads as soft deformation, meaning the geometry bends in response to the impact rather than swapping to a pre-authored damaged version.
That is a genuine technical difference, not a visual preference. GTA 5 used preset damage states, and a car in that game moves between a small number of authored conditions rather than deforming continuously.
Why GTA 5 used preset damage, and why that constraint is gone
This is the part most coverage skips. GTA 5 shipped on hardware from 2006. Preset damage states are cheap: a handful of meshes, no continuous deformation maths, no per-panel state to track across dozens of vehicles on screen. It was the right call for the machine it had to run on.
That constraint does not exist on a PS5. The interesting question is therefore not whether Rockstar could do soft deformation now, because it obviously can, but how far it chose to take it across an entire open world rather than in a scripted trailer shot.
Gellhorn International Raceway
The Extended Look included a race at Gellhorn International Raceway. A racetrack sequence is useful because it shows cars at speed on a surface designed for it, and it is limited for exactly the same reason: a track is the most forgiving environment a handling model will ever face.
Watch it for what the cars do under load and how weight transfers. Do not read it as a statement about how the same car behaves on a wet street with traffic, because the footage does not contain that.
Why the getaway scene proves less than it looks like
A getaway sequence is edited, framed and driven by someone who knows what happens next. Every trailer getaway in the history of this series has looked like it demonstrated the handling model, and none of them ever did.
The video is explicit about this. A scene assembled to look good is evidence about editing, and treating it as evidence about physics is the most common mistake in this entire discussion.
The August leak clips stay at Reported
Press reporting on the August 2026 leak clips described handling behaviour, and that reporting sits at the Reported tier here: named outlets, attributed, not proven, and describing material from an unknown build. No leaked footage is shown in the video or on this page.
Mixing that into the confirmed pile would make the case look stronger than it is, which is the opposite of the point.
What the video covers
- The confirmed Rockstar interviews, and what they do not settle
- The in-engine door deformation from Trailer 1
- The race at Gellhorn International Raceway in the Extended Look
- Why the getaway scene proves less than it looks like it proves
- Why GTA 5 used preset damage instead of soft deformation, and why that constraint is gone on PS5
- Press reporting on the August leak clips, kept at Reported
What we do not know yet
- The handling model itself, which no released footage settles
- How far soft deformation extends across the open world rather than in trailer shots
- Whether driving feel changes between the two protagonists or vehicle classes
Sources
Evidence tier for the racing drivers and the Rob Nelson comments: confirmed, on the record. For the door deformation and the raceway footage: observed. For the August leak clips: reported. The conclusion about which way it leans is flagged in the video as a read, not a finding. See how this site grades evidence.
- IGN, Rob Nelson interview: the narrower streets and the comment that driving takes a bit of skill
- GamingBolt: reporting on the narrower streets, traffic density and the increased difficulty of driving
- Grand Theft Auto VI Trailer 1 and the Extended Look, Rockstar Games: the in-engine door deformation and the Gellhorn International Raceway sequence
Both GTA 6 leaks side by side, and how to tell a real one from a fake: GTA 6 leaks, explained.
GTA 6 
