Cohere 发布 Parse 5,可以从复杂的文档中高效地提取多模态信息
Cohere 公司发布名为 Parse 5 的新工具,旨在从复杂文档中高效提取多模态信息。该工具支持处理文本、图像及音频等多种格式内容,能够自动识别并关联不同模态数据间的语义关系。Parse 5 采用先进的注意力机制与检索增强生成技术,显著提升了信息抽取的准确率与速度。目前该工具已开源,开发者可基于其 API 构建定制化文档分析应用,适用于法律审查、医疗记录分析及学术研究等场景。
EVENT DOSSIER
On September 8, 2026, Cohere officially released a new tool named Parse 5. This tool is designed to efficiently extract multimodal information from complex documents, supporting processing of text, images, and audio, as well as automatically identifying and correlating semantic relationships between different modal data. Parse 5 utilizes advanced attention mechanisms and retrieval-enhanced generation techniques, significantly improving the accuracy and speed of information extraction. Currently, this tool is open-source, and developers can build customized document analysis applications based on its API, making it suitable for legal review, medical record analysis, and academic research.
Cohere 公司发布名为 Parse 5 的新工具,旨在从复杂文档中高效提取多模态信息。该工具支持处理文本、图像及音频等多种格式内容,能够自动识别并关联不同模态数据间的语义关系。Parse 5 采用先进的注意力机制与检索增强生成技术,显著提升了信息抽取的准确率与速度。目前该工具已开源,开发者可基于其 API 构建定制化文档分析应用,适用于法律审查、医疗记录分析及学术研究等场景。