Home: Motoring > Tesla Launches FSD v15 in the U.S., Musk Says Real-World View Uses Only Half the Compute of One AI4 TRIP Accelerator

Tesla Launches FSD v15 in the U.S., Musk Says Real-World View Uses Only Half the Compute of One AI4 TRIP Accelerator

From:Internet Info Agency 2026-11-22 12:41:45

On November 22, 2026, Tesla announced the launch of FSD v15 in the U.S. market, introducing a new visualization interface called "Tesla Reality View." Musk stated in a post on November 22, 2026, that FSD v15's "Tesla Reality View" is generated by real-time AI video without the need for a game engine, and uses only half the compute of a single TRIP accelerator on the AI4 chip. According to Tesla's release notes, "Tesla Reality View" uses the in-vehicle AI computer to generate in real time a virtual overhead view of the surrounding road environment. The interface supports full-screen, map, or split-screen views, and users can tap a target location on the visualization to direct the vehicle there. According to prior disclosures by Ashok Elluswamy, Tesla's head of AI, the neural network parameter count of FSD v15 has expanded from approximately 1 billion to approximately 10 billion. Tesla says the upgraded vision encoder is shared across FSD, Robotaxi, and the Optimus humanoid robot, and is trained on millions of internet videos. FSD v15 runs the same software stack as Tesla's Robotaxi fleet. Elluswamy disclosed during the Q2 2026 earnings call that Robotaxis running early v15 versions had achieved over 380,000 miles of unsupervised driving across six cities in two states, with no major accidents during that period. The AI4 referred to in the text is Tesla's fourth-generation autonomous driving hardware platform, HW4. Since FSD V13, Tesla has changed the naming of the HW4 chip platform to AI4. The platform offers roughly 5x the compute of HW3. According to previously leaked specifications, HW4's CPU cores increased from 12 to 20, with a maximum frequency of 2.35 GHz; its TRIP deep neural network accelerator was upgraded from dual-core to tri-core, with maximum frequency raised from 2 GHz to 2.2 GHz, and estimated total compute of approximately 300-500 TOPS.

Editor:NewsAssistant