Answers, not vibes
Does FSD handle it?
Pick the situation you actually care about — snow, roundabouts, construction, school buses — and see every filmed moment of it across seven years of DirtyTesla footage. Every clip links to the exact second.
14 scenarios
How is FSD at left turns?Every filmed left turn, with the unprotected ones — no green arrow, oncoming traffic, a judgement call about the gap — as the hard case inside it.1016 filmed moments · 319 videos →Is FSD good on the highway?Merging, lane selection, exit timing, and the passing-lane camping that owners complain about most.906 filmed moments · 208 videos →Can FSD park itself?Parking lots are unmapped, unmarked, and full of pedestrians — the place FSD behaves least like it does on the road.868 filmed moments · 227 videos →How does FSD react to pedestrians?Crosswalks, jaywalkers, cyclists, and the downtown Ann Arbor foot traffic that shows up in hundreds of clips.865 filmed moments · 251 videos →Does Tesla FSD work in snow?Michigan winters are the channel's home turf, so snow is one of the most heavily filmed conditions in the archive.602 filmed moments · 51 videos →How does FSD handle heavy rain?Rain degrades the camera-only stack in ways owners feel immediately — wiper behaviour, spray, standing water, and the occasional take-over-immediately alert.476 filmed moments · 69 videos →Does FSD understand construction zones?Cones, shifted lanes, flaggers, and temporary markings — where map data stops helping and the vision stack is on its own.328 filmed moments · 162 videos →Is FSD worse at night?Darkness, oncoming headlights, and unlit rural roads are a different problem than the daytime footage most reviews show.294 filmed moments · 30 videos →Does FSD work on dirt roads?No lane lines, soft edges, and drop-offs. Rural Michigan gave this archive an unusual amount of unmarked-road footage.261 filmed moments · 133 videos →Can FSD handle a roundabout?A long-running sore point with a genuinely visible arc across versions — worth watching in order.254 filmed moments · 127 videos →Does FSD slow down for speed bumps?The comfort problems that do not show up in safety statistics but decide whether you actually use FSD every day.133 filmed moments · 103 videos →Is phantom braking fixed?The complaint that outlived more FSD versions than any other. Braking for a real obstacle is excluded — this is only the braking nobody could explain.132 filmed moments · 105 videos →What does FSD do in fog?The rarest filmed condition in the archive, and the one where the camera-only argument gets tested hardest.68 filmed moments · 20 videos →Does FSD stop for school buses and emergency vehicles?Legally mandated behaviour with real consequences — and one of the clearest pass/fail tests in the archive.48 filmed moments · 42 videos →
No scenarios match that search. Try a condition, road type, or driving behavior.
Coverage matrix
Filmed moments per scenario per software era. Dim cells are thin samples (fewer than 3 videos); a dot means nothing filmed in that condition on that build — which is not the same as a problem being fixed.
| Version | Left turns | Highway merges | Parking lots | Pedestrians | Snow | Rain | Construction zones | Night driving | Dirt | Roundabouts | Speed bumps | Phantom braking | Fog | School buses |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| v14.3 | 18 | 18 | 133 | 47 | · | 50 | 19 | · | 5 | 6 | 6 | 9 | 1 | 1 |
| v14.2 | 12 | 12 | 78 | 24 | 49 | · | 2 | · | · | 11 | 6 | 1 | · | 1 |
| v14.1 | 17 | 8 | 70 | 34 | 16 | 25 | 17 | 27 | 7 | 6 | 1 | 12 | · | 1 |
| v14 | 1 | · | 1 | 2 | · | · | · | 6 | · | · | · | · | · | 2 |
| v13.2 | 43 | 19 | 77 | 36 | 63 | 20 | 14 | · | 6 | 19 | 7 | · | 1 | 3 |
| v12.5 | 53 | 60 | 31 | 44 | 18 | 21 | 19 | 50 | 17 | 6 | 9 | 5 | · | 4 |
| v12.4 | 11 | 10 | 7 | 9 | · | · | 5 | · | 4 | 3 | 2 | 1 | 1 | 2 |
| v12.3 | 26 | 6 | 21 | 31 | 25 | 21 | 10 | · | 4 | 5 | 3 | 1 | · | · |
| v12.1 | 1 | · | · | 1 | · | · | · | · | · | · | 2 | · | · | · |
| v12 | 5 | 2 | 6 | 11 | 26 | 1 | 1 | · | · | · | · | · | · | 1 |
| v11.4 | 97 | 79 | 18 | 106 | 18 | 41 | 46 | 20 | 11 | 30 | 7 | 9 | · | 2 |
| v11.3 | 51 | 31 | 17 | 43 | 16 | 19 | 15 | · | 5 | 12 | 4 | 3 | 1 | 1 |
| v11 | 5 | · | · | 9 | · | · | 2 | · | 23 | · | · | · | · | · |
| v10.69 | 149 | 28 | 50 | 103 | 21 | 36 | 31 | 7 | 18 | 38 | 8 | 15 | 17 | 8 |
| v10 | 221 | 27 | 52 | 197 | 121 | 38 | 48 | 126 | 27 | 36 | 18 | 27 | 30 | 3 |
| v9 | 43 | 8 | 20 | 18 | · | · | 5 | 22 | 14 | 5 | 3 | 4 | 1 | · |
| v8 | 51 | 10 | 40 | 28 | 53 | · | 4 | · | 6 | 5 | 5 | 4 | · | 1 |
| v14 Lite | 1 | · | 7 | 3 | · | · | 1 | · | 1 | · | 2 | 1 | · | · |
How scenarios are tagged: each moment is matched from scenario-specific
extracted text fields, plus the parent video's weather and drive type where relevant.
It finds what was described on camera — a condition that went unmentioned will not appear.
Counts are observations, never rates. Full methodology.