LlamaIndex 发布 ParseBench-2:59 个文档解析模型横评
做 RAG 的朋友可以看下 LlamaIndex 这份 ParseBench-2 榜单,59 个解析模型横评,开源阵营里 KDL 和 Infinity-Parser2 排最前。选文档解析模型不用再瞎试了。
LlamaIndex 团队在 Hugging Face 上发布文档解析基准 ParseBench-2,共评测 59 个模型,其中 20 个为开源权重模型。排名靠前的开源模型包括 KDL、Infinity-Parser2、LightOnOCR-3 和 Chandra 2。该基准上线一周已累计 4.2 万+ 月下载量,数据集托管在 Hugging Face,评测详情可访问 parsebench.ai 查看。
ParseBench-2 专注于文档解析任务,即把 PDF、扫描件等原始文档转成结构化文本,这是 RAG 管道中常被忽视但直接影响检索质量的一环。榜单同时覆盖开源与闭源模型,方便开发者按解析精度与部署成本权衡选型。
文档解析质量决定 RAG 效果上限,此前缺乏统一的公开评测,这份榜单可直接作为选型参考。对自建文档处理管道的企业,开源权重模型已能与闭源方案正面比较。
We've benchmarked 20 open-weight models on @huggingface (and 59 models total!) on ParseBench-2 for document parsing tasks. The top open-weight models are KDL, Infinity-Parser2, LightOnOCR-3, Chandra 2. It's been a week and we've hit 42k+ monthly downloads 📈 come check it out! Dataset: huggingface.co/datasets/llama… d Site: parsebench.ai X📊 3 ⚡
This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include ParseBench-2, LlamaIndex, Hugging Face, RAG.
Source attribution: curated from Jerry Liu (@jerryjliu0).