XPENG releases TuringViT for smart driving and humanoid robots

XPeng has released TuringViT, a vision encoder for vision-language models used in autonomous driving, cockpit systems, and its IRON humanoid robot. The model achieves 3x throughput of competing vision transformers at 1536×1536 resolution and 83.6% average zero-shot benchmark performance on 850 million training pairs.

This Wire brief sits within Fusion42's coverage of Computer Vision, AI Frontier Models and Autonomous Vehicles.