做PDF智能体的话，这个基准和工具值得看。通用模型定位不行，LlamaParse能精确到词级框，长文档也不掉链子。
LlamaIndex发布ExtractBench基准，专门评估文档提取的视觉定位能力。测试显示，通用视觉模型和编码智能体在返回证据时得分为零，而专用VLM文档工具LlamaParse表现领先。LlamaExtract Agentic Plus在页面级F1达84.9%，词级46.4%，长文档上保持87.1%，其他系统在长文档上跌至0%。基准要求字段值和引用都正确才算分，词级框IoU需达0.5。
If you’re building an “agent over your PDFs” in 2026, one of the most underrated things you can do i...
If you’re building an “agent over your PDFs” in 2026, one of the most underrated things you can do is to provide exact grounding back to the source document for any agentic answers. Specifically, showing the exact word, line, or region that the answer came from. Frontier vision models do a terrible job at this. They have a hard time predicting exact bounding boxes over region, tying it back to the source answer, and surfacing it as context in a cost-effective and accurate manner. Dedicated VLM-based document OCR/extraction tools like LlamaParse are a lot better at visual grounding and specifically tying it to extracted outputs. One of our main evaluation areas in ExtractBench, our recently released real-world benchmark, is to evaluate the visual grounding capabilities of document extraction. ExtractBench: extractbench.ai If you want to give LlamaParse a shot, come check it out: cloud.llamaindex.ai Your browser does not support the video tag. 🔗 View on Twitter LlamaIndex 🦙 @llama_index Most document extraction APIs can't tell you where a value came from. For ExtractBench, we scored grounding strictly: a field only counts if the value AND its citation are correct, word-level box at IoU 0.5. A perfect box around a wrong value earns nothing. Results: VLMs and coding agents return no evidence at all — zero at both levels. Among systems that do return boxes, the best word-level F1 is still under 50%. And grounding collapses with length: one specialized API goes from 61.7% page-level on short docs to 0.0% on long ones. LlamaExtract Agentic Plus leads at both levels — 84.9% page-level, 46.4% word-level — and holds at 87.1% on long documents where others hit zero. Every extracted value should come with receipts. ExtractBench now gives the field a baseline to track it 👇 Blog: lnkd.in/gNm97fXp 6 Paper: lnkd.in/euAfScWx F Your browser does not support the video tag. 🔗 View on Twitter 🔗 View Quoted Tweet 💬 0 🔄 0 ❤️ 6 👀 1168 📊 2 ⚡
- LlamaIndex08-17 15:59原文