the-decoder

Deepseek OCR 2 cuts visual tokens by 80% and outperforms Gemini 3 Pro on document parsing

Deepseek has released Deepseek OCR 2, a vision encoder that processes image information based on content context, requiring only 256 to 1,120 tokens per image—significantly fewer than comparable ...
腾讯网

DeepSeek OCR 2发布：开源OCR领域迎来强劲选手

DeepSeek悄然在Hugging Face上线了DeepSeek-OCR-2（huggingface.co/deepseek-ai/DeepSeek-OCR-2），这款专注于文档识别的开源模型引发了社区热议 ...
TechNode

DeepSeek releases OCR 2 with new visual encoding architecture, targeting more human-like ...

Chinese AI startup DeepSeek on Tuesday released a research paper and open-sourced its latest optical character recognition (OCR) model, DeepSeek-OCR 2, aiming to improve how machines interpret and ...
scmp.com

DeepSeek taps Alibaba open-source AI technology to boost OCR performance

Chinese artificial intelligence start-up DeepSeek on Tuesday unveiled an upgraded version of its optical character recognition (OCR) model, incorporating an Alibaba Cloud-developed open-source system ...
腾讯网

DeepSeek-OCR 2发布：让AI像人一样“读懂”复杂文档

凤凰网科技讯 1月27日，DeepSeek团队发布《DeepSeek-OCR 2: Visual Causal Flow》论文，并开源DeepSeek-OCR 2模型。 该模型采用创新的DeepEncoder V2新型编码器结构，它能够根据图像语义动态调整视觉信息的处理顺序，使模型在进行文字识别前先对视觉内容进行智能排序。
搜狐

刚刚，DeepSeek 发布 OCR 2

刚刚，DeepSeek 发布了 新模型 DeepSeek-OCR 2，采用创新的DeepEncoder V2方法，让AI能够根据图像的含义动态重排图像的各个部分，更接近人类的视觉编码逻辑。在具体实现上，DeepSeek 团队在论文中称采用了 Qwen2-0.5B 来实例化这一架构。 如果说去年 10 月 DeepSeek-OCR 的发布 ...
搜狐

DeepSeek 突发OCR 2，采用基于Qwen的新架构

阿里半夜刚发完旗舰模型，这边 DeepSeek坐不住了，突然发布更新了。 刚刚，DeepSeek 发布了 新模型 DeepSeek-OCR 2，采用创新的DeepEncoder V2方法，让AI能够根据图像的含义动态重排图像的各个部分，更接近人类的视觉编码逻辑。在具体实现上，DeepSeek 团队在论文中称 ...
Cleveland.com

Cuyahoga County upgraded to Level 2 snow emergency as dangerous conditions worsen

CLEVELAND, Ohio - Cuyahoga County Sheriff Harold Pretel has elevated the county to a Level 2 Snow Emergency due to heavy snowfall and dangerous driving conditions, according to a news release.