Give DeepSeek V4 a 3B vision front-end — it works today

给性价比最高的推理引擎装一双眼睛

DeepSeek V4's API rejects image input, but Liquid AI's new LFM2.5-VL-3B can serve as a perception layer. The author tested it zero-shot on garage camera data — 94.3% accuracy for door open/close, 1.5s locally. Perception runs on-device, images never leave, only structured text hits the cloud for reasoning. The latency, privacy, and cost advantages of this split architecture hold even if DeepSeek adds native vision later. One gap: extracting a reliable confidence signal from natural-language output — the post doesn't detail a calibration method.

Why it matters: The author built a perception frontend for DeepSeek V4 using Liquid's LFM2.5-VL-3B, tested it on a real garage camera with 94.3% accuracy and 1.5s latency — concrete data, clear architecture. Also traces DeepSeek's vision research history (VL2, the retracted Thinking with Visu...