什么是 Jina AI？

Jina AI发布视觉文档解析模型jina-ocr-v1，内置推测解码技术，支持PDF/扫描件等格式转markdown，参数量达34亿。

Jina AI发布的jina-ocr-v1在25种语言支持和NVIDIA L4 GPU优化上表现突出，OmniDocBench和olmOCR-Bench得分分别为91.14和83.4。

Jina AI发布的新版重排序模型jina-reranker-v3.5在参数量更少的情况下，于BEIR数据集上表现优于竞品。

Jina AI发布的jina-reranker-v3.5在四个基准测试中表现优异，参数量仅为竞品的七分之一。

Jina AI 推出支持多模态搜索的本地化工具，适用于 Mac 平台，无需 Python。

Jina AI releases the v5-omni model, retaining the v5-text backbone and adding pretrained vision and audio encoders connected through small trainable projectors.

GenAI turns raw logs into actionable insights—Elastic’s integration of Jina AI models and Search AI reduces MTTR by up to 60% in real-world deployments.

Jina AI发布v3.5 reranker模型，在参数减少7倍的情况下，BEIR数据集表现优于Qwen3-Reranker-4B。

Elastic加入Open Secure AI Alliance，推动AI安全与开放创新。该联盟旨在通过共享工具和研究提升AI安全性，Elastic将贡献开源安全和AI检测技术。

Elastic Stack 9.3.5 发布，新增 Piped Query Language 和 Jina AI 搜索模型支持，提升数据处理效率。

Jina AI has released v5-omni, the first universal embedding model supporting text, images, audio, and video, available in two sizes with Matryoshka truncation support.

Elastic Stack 9.4.1 has been released, adding features such as Context Engineering and Jina AI search models.

v5-omni is now available on Elastic Inference Service, HuggingFace, and Jina API, providing more resources for learning.

Jina AI将在EMNLP2026举办关于搜索技术的BoF会议，邀请提交嵌入、重排序器和小型语言模型相关研究。

Jina AI 宣布发布 OCR 模型 jina-ocr-v1，但未提供技术细节或使用价值说明。

This page is primarily site navigation structure without specific article body. Although it mentions using Search AI for logs and Jina AI models, it lacks mechanism depth and low information density.