Tesla rolls out FSD v15: 25x better safety-critical handling, shared Optimus vision encoder
EricBuess · x · 2026-11-22
Tesla's FSD v15 is rolling out, upgrading every part of the end-to-end driving network:
- Upgraded vision encoder shared between FSD, Robotaxi and Optimus, trained on millions of internet videos — improving critical intervention rate 25x
- Tesla Reality View: replaces all explicit model signals with a single end-to-end live visualization of what FSD plans, using less compute
- End-to-end navigation learned from fleet route preferences for trip start/end, expanding to full routes later
- Other gains: personalized parking selection, better driver monitoring with hats/sunglasses, improved detection of signs/animals/debris/potholes/standing water, more natural school-zone and bad-weather speed handling, slowing for speed cameras, and a Bike Rack Mode
Related event: Tesla Rolls Out FSD v15 with 25x Improvement in Safety-Critical Scenarios(4 posts)→
More from Embodied
- Auki founder at Humanoid Hub: physical AI should run physical businesses, not just move robots — broodsugar · 2026-11-22
- Chart: Tesla FSD hits 438K weekly Austin rides, overtaking Waymo's 406K within months — MatthewChang · 2026-11-22
- Figure CEO shares photo of humanoid robot at Archer event — adcock_brett · 2026-11-22
- Open-source sub-$100 quadruped and humanoids run RL policies in your browser via MuJoCo — RemiCadene · 2026-11-22
- Teacher asks if a 6GB RTX 3060 can run a local Claude math-grading agent — beekersavant · 2026-11-22
- Sony's ping-pong robot Ace beats world No.26 pro with 20.2ms reaction time — tak3sh8 · 2026-11-22