精选理由
别信“大模型能吃 OCR”。LlamaIndex 用三代 GPT 数据反驳：成本涨 4 倍，精度还落后。
LlamaIndex 对比三代 GPT 模型的文档解析表现，发现精确度只提升约 24 个百分点，但每页成本涨了 4 倍。即便最新一代前沿模型，在 OCR 任务上仍落后于专用解析器。作者认为“大模型会吃掉 OCR”的说法缺乏数据支撑，性能差距还会随成本放大。
原文 · LlamaIndex
"OCR is just a feature now. Frontier models will eat it." We hear this constantly. The data says ot...
"OCR is just a feature now. Frontier models will eat it." We hear this constantly. The data says otherwise. Across three GPT generations, parsing accuracy gained ~24 points, while cost per page 4x'd. And the newest frontier models still trail specialized parsers. Read more below on why the gap persists (and compounds) 👇️ llamaindex.ai/blog/document-… b 💬 0 🔄 0 ❤️ 1 👀 416 ⚡