According to @karpathy, a new HW4 Tesla Model X running FSD v13 completed a smooth, confident highway and city route that handled lane centering, construction detours, tricky left turns, four-way stops, bus overtakes, dense merges, parking, and ended as a perfect drive with no notes, indicating a markedly better experience than HW3. Source: Andrej Karpathy on X, Nov 12, 2025. According to @karpathy, the results reflect FSD v13 on HW4 because his car has not yet received v14, providing a current field-performance reference for traders tracking Tesla’s autonomy progress. Source: Andrej Karpathy on X, Nov 12, 2025. According to @karpathy, progress is driven by an end-to-end long-context neural network that processes surround video at 60 Hz with multimodal sensor streams over roughly 30 seconds, with technical hints attributed to Ashok Elluswamy’s ICCV25 talk. Source: Andrej Karpathy on X, Nov 12, 2025; Ashok Elluswamy on X (ICCV25 talk referenced by Karpathy). According to @karpathy, this firsthand report underscores a material performance gap in favor of HW4 versus HW3 for FSD v13, a datapoint TSLA-focused traders can use when evaluating hardware-driven capability differences in Tesla’s fleet. Source: Andrej Karpathy on X, Nov 12, 2025. According to @karpathy, no cryptocurrencies, blockchain integrations, or digital assets are mentioned in this report, implying no direct crypto market linkage in the update. Source: Andrej Karpathy on X, Nov 12, 2025.