{"@context":"https://schema.org","@type":"NewsArticle","generatedAt":"2026-07-28T06:20:51.496Z","headline":"Datalab Marker v2 发布：三模式管道在 olmOCR-bench 上达 76.0，速度超 MinerU 5 倍以上","description":"Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 倍以上，同时在准确率和速度上均优于 Docling。","url":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/","mainEntityOfPage":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/","datePublished":"2026-07-25T04:42:47.000Z","dateModified":"2026-07-25T04:42:47.000Z","inLanguage":"zh-CN","publisher":{"@type":"NewsMediaOrganization","name":"Aioga","url":"https://www.aioga.com"},"citation":["https://www.marktechpost.com/2026/07/24/datalab-marker-v2-vs-mineru-docling-and-liteparse-benchmark-breakdown","https://aihot.virxact.com/items/cmrzwd6wi01nmrop1ltzp5nza"],"canonicalUrl":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/","directAnswer":{"@type":"Answer","text":"Aioga 编辑摘要：Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 Aioga 将其归入「产品更新」方向，重点关注它对真实使用和行业竞争的影响。","url":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/","dateCreated":"2026-07-25T04:42:47.000Z","author":{"@type":"Organization","@id":"https://www.aioga.com/authors/aioga-editorial/#editorial-team","name":"Aioga Editorial Team","url":"https://www.aioga.com/authors/aioga-editorial/"}},"evidence":[{"@type":"CreativeWork","name":"marktechpost.com source article","url":"https://www.marktechpost.com/2026/07/24/datalab-marker-v2-vs-mineru-docling-and-liteparse-benchmark-breakdown","datePublished":"2026-07-25T04:42:47.000Z","provider":{"@type":"Organization","name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/24/datalab-marker-v2-vs-mineru-docling-and-liteparse-benchmark-breakdown"}},{"@type":"CreativeWork","name":"AIHot archive record","url":"https://aihot.virxact.com/items/cmrzwd6wi01nmrop1ltzp5nza","datePublished":"2026-07-25T04:42:47.000Z","provider":{"@type":"Organization","name":"AIHot","url":"https://aihot.virxact.com/items/cmrzwd6wi01nmrop1ltzp5nza"}}],"aggregationSource":"MarkTechPost（RSS）","originalPublisher":{"name":"marktechpost.com","url":"https://www.marktechpost.com/2026/07/24/datalab-marker-v2-vs-mineru-docling-and-liteparse-benchmark-breakdown"},"article":{"id":"cmrzwd6wi01nmrop1ltzp5nza","slug":"cmrzwd6wi01nmrop1ltzp5nza","url":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/","title":"Datalab Marker v2 发布：三模式管道在 olmOCR-bench 上达 76.0，速度超 MinerU 5 倍以上","title_en":"Datalab Marker v2 vs MinerU， Docling， and Liteparse： Benchmark Breakdown","summary":"Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 倍以上，同时在准确率和速度上均优于 Docling。","source":"MarkTechPost（RSS）","sourceUrl":"https://www.marktechpost.com/2026/07/24/datalab-marker-v2-vs-mineru-docling-and-liteparse-benchmark-breakdown","aiHotUrl":"https://aihot.virxact.com/items/cmrzwd6wi01nmrop1ltzp5nza","publishedAt":"2026-07-25T04:42:47.000Z","category":"产品更新","score":55,"selected":false,"articleBody":["Datalab has released Marker 2：https://pxllnk.co/c1pzvpb , a full rewrite of its open source document conversion pipeline. Marker converts PDF, image, PPTX, DOCX, XLSX, HTML, and EPUB files into markdown, JSON, HTML, or chunks. The Datalab team rebuilt it around three components shipped over the preceding months: Surya OCR 2 , a 20M-param fast layout model , and a rebuilt pdftext that is 3× faster than the previous one.","The main result comes from olmOCR-bench：https://github.com/allenai/olmocr/tree/main/olmocr/bench, a third-party benchmark from Allen AI. Marker 2’s ：https://pxllnk.co/c1pzvpbbalanced mode scores 76.0% overall and 83.5% on born-digital PDFs. It sustains 2.9 pages per second on a single B200 GPU. That is over 5× the throughput of MinerU’s pipeline backend, which scores 72.7% at 0.54 pages per second. Docling scores 50.3% at 2.1 pages per second on the same harness.","Marker 2 ：https://pxllnk.co/c1pzvpbexposes three conversion paths instead of one:","Mode is now device-aware by default : balanced on GPU, fast on CPU/MPS, overridable with –mode. Full CPU support is the second structural change. fast –disable_ocr needs no GPU and no inference server, and the 20M layout model still reads columns, tables and headers on CPU.","The third change is architectural, and it is the one that produces the throughput numbers. Many thin CPU workers share a single Surya inference server. The parent process budgets VLM concurrency across them, so throughput scales with server capacity rather than per-process VRAM. Datalab reports that balanced mode sustains ~2.9 pg/s against a ~0.3 pg/s single-stream rate on the same hardware.","Breaking changes are worth flagging before an upgrade. Python 3.10+ is now required. Packaging moved from Poetry to uv , with hatchling as the build backend, though pip install marker-pdf is unchanged. The structured-extraction converter and extractors were removed; Datalab points users to the hosted API or a –use_llm workflow instead.","The scoring benchmark is olmOCR-bench from Ai2: 1,403 PDFs with roughly 8,400 pass/fail unit tests covering math rendering, table structure, reading order, headers and footers, and old scans. The overall score is the macro-average across the 8 categories, computed with the official olmOCR-bench checker. Throughput is sustained concurrent pg/s on one B200 host, not single-stream latency.","A note on provenance. olmOCR-bench is a third-party benchmark from Ai2, but every score and throughput figure below comes from Datalab’s own runs. All of them are reproducible through the open harness in the Marker repository, which ships competitor runners for MinerU, Docling and LiteParse alongside Marker’s own.","These numbers also reflect one benchmark’s document mix measured on a single hardware setup, so results on your own documents may differ. Teams evaluating these systems should run the harness against their own corpus, which is the only way to know how the four rank on the documents they actually process.","MinerU’s pipeline backend is the closest architectural match. Both read the PDF text layer and OCR selectively. On overall score, Marker balanced leads 76.0 to 72.7. On born-digital documents the two are effectively tied: 83.5 against 83.3.","The separation is throughput. Marker balanced sustains 2.9 pg/s against MinerU’s 0.54 pg/s, a 5.4× gap at a higher score. Marker fast sustains 7.4 pg/s, roughly 13.7× MinerU’s pipeline rate, but scores 6.1 points below MinerU to do it.","MinerU also ships a VLM backend , which Datalab states scores higher than its pipeline backend. That backend is a full-page-VLM approach and is not in this table. AI teams evaluating MinerU should benchmark that path separately.","Docling is the widest margin among the GPU pipelines. Marker balanced leads 76.0 to 50.3 overall and 83.5 to 64.0 on born-digital, while also running faster: 2.9 pg/s against 2.1 pg/s. Datalab notes Docling was run on its default pipeline, which uses the text layer for born-digital pages and OCR for image regions.","Docling’s counterweight is governance and format breadth, not accuracy. The codebase is MIT-licensed , it originated at IBM Research, and it is hosted as a project in the LF AI & Data Foundation . Its input list also extends past documents into audio and email formats.","LiteParse, from the LlamaIndex team, is a Rust document parser. It does not compete on the same axis. On CPU it scores 22.4 overall and 20.4 with OCR off, against Marker’s CPU-only 43.6.","But LiteParse with OCR disabled reports 1721 pg/s — roughly 73× Marker’s CPU mode, which is the tradeoff. Marker’s fast –disable_ocr runs a 20M layout model on CPU and still recovers structure, which is why it more than doubles a plain text dump’s score. LiteParse has no layout model and collapses on anything non-linear.","The Datalab team emphasizes that Marker is designed as a pipeline rather than a VLM, clarifying that these are distinct tools. In this evaluation, their hosted Chandra 2 scores 85.8, while Gemini Flash 3.5 via API scores 76.4. Datalab’s Chandra repository also positions Ai2’s olmOCR 2 at 82.4 and dots.ocr 1.5 at 83.9 within a separate table. For scans, math-heavy pages, and achieving top accuracy, the VLM tier remains superior to all listed pipelines.","Marker’s balanced mode narrows the performance gap to just 0.4 points behind Gemini Flash 3.5 overall, and it even outperforms it on born-digital documents by a margin of 83.5 to 79.1—without requiring a per-page API call.","The mode you pick changes the failure profile, not just the score. Each row is one olmOCR-bench category, scored across all three modes. Math is the sharp edge: fast mode reads equations from the PDF text layer instead of VLM-OCRing them, so arXiv math falls from 83.9 to 23.4, and –disable_ocr scores 0.0 there by design. Outside the two math categories, old scans is the weakest split in every mode, topping out at 43.2.","This is where the four systems diverge most for commercial teams:","Score alone does not pick the tool. Corpus type, hardware, licensing band and output format decide it. Try the interactive picker below to filter ten deployment scenarios by constraint and by tool, and see which parser fits your use case.","Links: GitHub repo：https://pxllnk.co/c1pzvpb | Release notes：https://github.com/datalab-to/marker/releases/tag/v2.0.0 | Blog post：https://www.datalab.to/blog/marker-2 | Announcement tweet：https://x.com/VikParuchuri/status/2079545884681830784","Asif Razzaq is the CEO of Marktechpost Media Inc.. As a visionary entrepreneur and engineer, Asif is committed to harnessing the potential of Artificial Intelligence for social good. His most recent endeavor is the launch of an Artificial Intelligence Media Platform, Marktechpost, which stands out for its in-depth coverage of machine learning and deep learning news that is both technically sound and easily understandable by a wide audience. The platform boasts of over 2 million monthly views, illustrating its popularity among audiences."],"articleImages":[{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/07/image-4.png","alt":"","afterParagraph":8,"url":"/media/articles/cmrzwd6wi01nmrop1ltzp5nza/d9cdc88d28f92dce.webp"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/07/image-6-1024x698.png","alt":"","afterParagraph":18,"url":"/media/articles/cmrzwd6wi01nmrop1ltzp5nza/4efca9f6b6c69348.webp"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2019/06/Screen-Shot-2021-09-14-at-9.02.24-AM-300x300.png","alt":"","afterParagraph":21,"url":"/media/articles/cmrzwd6wi01nmrop1ltzp5nza/787a6d54564e8e19.webp"},{"sourceUrl":"https://www.marktechpost.com/wp-content/uploads/2026/07/blog6171-1-100x70.png","alt":"Designing High-Performance GPU Kernels with TileLang: Tensor-Core GEMM, Fused Softmax, FlashAttention, and Autotuning","afterParagraph":22,"url":"/media/articles/cmrzwd6wi01nmrop1ltzp5nza/80925b48ffa65ad2.webp"}],"mediaStatus":"ok","articleBodyZh":["Datalab 发布了 Marker 2：https://pxllnk.co/c1pzvpb，这是其开源文档转换管道的完全重写版本。Marker 可以将 PDF、图像、PPTX、DOCX、XLSX、HTML 和 EPUB 文件转换为 Markdown、JSON、HTML 或分块格式。Datalab 团队围绕之前几个月发布的三个组件对其进行了重建：Surya OCR 2、一款 2000 万参数的快速布局模型，以及重建后的 pdftext，其速度是以前的 3 倍。","主要结果来自 olmOCR-bench：https://github.com/allenai/olmocr/tree/main/olmocr/bench，这是 Allen AI 提供的第三方基准测试。Marker 2 的：https://pxllnk.co/c1pzvpbbalanced 模式总体得分为 76.0%，在原生数字 PDF 上得分为 83.5%。在单个 B200 GPU 上可维持每秒 2.9 页的速度。这是 MinerU 的管道后端吞吐量（每秒 0.54 页，总得分 72.7%）的 5 倍以上。在同一测试环境下，Docling 的得分为 50.3%，速度为每秒 2.1 页。","Marker 2：https://pxllnk.co/c1pzvpb 提供了三种转换路径，而不是单一路径：","模式现在默认会根据设备进行选择：在 GPU 上为 balanced，在 CPU/MPS 上为 fast，可以通过 --mode 覆盖。全面的 CPU 支持是第二个结构性变化。fast --disable_ocr 不需要 GPU 或推理服务器，而 2000 万参数的布局模型仍然可以在 CPU 上读取列、表格和标题。","第三个变化是架构上的，也是产生吞吐量数据的原因。多个轻量 CPU 工作线程共享单个 Surya 推理服务器。父进程会在它们之间分配 VLM 并发资源，因此吞吐量随服务器容量而不是每个进程的显存而扩展。Datalab 报告称，balanced 模式在相同硬件上可维持约每秒 2.9 页，而单流速度约为每秒 0.3 页。","在升级前值得注意的重大变更。现在需要 Python 3.10 及以上版本。打包方式从 Poetry 转移到 uv，使用 hatchling 作为构建后端，但 pip install marker-pdf 保持不变。结构化提取转换器和提取器已被移除；Datalab 建议用户使用托管 API 或 --use_llm 工作流。","评分基准是来自 Ai2 的 olmOCR-bench：1,403 个 PDF，约 8,400 个通过/未通过的单元测试，涵盖数学渲染、表结构、读取顺序、页眉页脚以及旧扫描件。总体得分是 8 个类别的宏平均值，使用官方 olmOCR-bench 检查器计算。吞吐量是单个 B200 主机上持续的并发页/秒，而非单流延迟。","关于来源的说明。olmOCR-bench 是 Ai2 的第三方基准，但以下所有得分和吞吐量数据均来自 Datalab 自身的运行。所有数据都可以通过 Marker 仓库中的开放测试框架重复获得，仓库中提供了 MinerU、Docling 和 LiteParse 的竞争对手运行器，以及 Marker 自身的运行器。","这些数字也反映了在单一硬件设置下对一个基准的文档组合的测量结果，因此在您自己的文档上可能会有所不同。评估这些系统的团队应对其自己的语料库运行测试框架，这是了解四种系统在实际处理文档时表现的唯一方法。","MinerU 的管线后端在架构上最为相似。两者都读取 PDF 文本层并选择性地进行 OCR。在总体得分上，Marker balanced 以 76.0 对 72.7 领先。在原生数字文档上，两者基本持平：83.5 对 83.3。","区别在于吞吐量。Marker balanced 可持续 2.9 页/秒，而 MinerU 为 0.54 页/秒，高得分下差距为 5.4 倍。Marker fast 可持续 7.4 页/秒，大约是 MinerU 管线速度的 13.7 倍，但得分低于 MinerU 6.1 分以实现该速度。","MinerU 还提供了一个 VLM 后端，Datalab 表示其得分高于管线后端。该后端采用全页 VLM 方法，未在本表中列出。评估 MinerU 的 AI 团队应单独对该路径进行基准测试。","Docling 在 GPU 管线中差距最大。Marker balanced 总得分以 76.0 对 50.3 领先，原生数字文档得分 83.5 对 64.0，同时运行速度更快：2.9 页/秒对 2.1 页/秒。Datalab 指出，Docling 在其默认管线上运行，该管线在原生数字页面使用文本层，在图像区域使用 OCR。","Docling 的支点是治理和格式广度，而非准确性。其代码库采用 MIT 许可证，起源于 IBM 研究院，并作为 LF AI & Data Foundation 的一个项目进行托管。其输入列表也扩展到了音频和电子邮件格式，而不仅限于文档。","LiteParse 来自 LlamaIndex 团队，是一个 Rust 文档解析器。它不在同一维度竞争。在 CPU 上其整体得分为 22.4，关闭 OCR 时为 20.4，而 Marker 的 CPU-only 模式得分为 43.6。","但禁用 OCR 的 LiteParse 报告速度为 1721 页/秒——约为 Marker CPU 模式的 73 倍，这就是权衡。Marker 快速的 --disable_ocr 在 CPU 上运行 2000 万布局模型仍能还原结构，这就是为什么其得分超过纯文本转储的两倍。LiteParse 没有布局模型，遇到任何非线性内容就会崩溃。","Datalab 团队强调，Marker 设计为管道而非 VLM，澄清了这些是不同的工具。在此次评估中，其托管的 Chandra 2 得分为 85.8，而通过 API 的 Gemini Flash 3.5 得分为 76.4。Datalab 的 Chandra 仓库还在单独的表格中将 Ai2 的 olmOCR 2 定位为 82.4，将 dots.ocr 1.5 定位为 83.9。对于扫描件、数学密集页以及实现顶级准确率，VLM 层级仍优于所有列出的管道工具。","Marker 的平衡模式将整体性能差距缩小到仅比 Gemini Flash 3.5 落后 0.4 分，并且在原生数字文档上的表现甚至优于其，得分 83.5 对 79.1——无需每页调用 API。","你选择的模式会改变失败模式，而不仅仅是得分。每行表示一个 olmOCR-bench 类别，在三种模式下进行评分。数学是尖锐点：快速模式从 PDF 文本层读取公式，而非进行 VLM-OCR，因此 arXiv 数学得分从 83.9 降至 23.4，--disable_ocr 在此处得分设计为 0.0。除两类数学外，旧扫描件是每种模式下最弱的一类，最高仅为 43.2。","这是四个系统在商业团队中分化最明显的地方：","仅凭分数无法选择工具。语料类型、硬件、许可类别和输出格式决定选择。尝试下面的交互式选择器，通过约束和工具筛选十种部署场景，看看哪个解析器适合你的使用场景。","链接：GitHub 仓库：https://pxllnk.co/c1pzvpb | 发布说明：https://github.com/datalab-to/marker/releases/tag/v2.0.0 | 博客文章：https://www.datalab.to/blog/marker-2 | 公告推文：https://x.com/VikParuchuri/status/2079545884681830784","Asif Razzaq 是 Marktechpost Media Inc. 的首席执行官。作为一位有远见的企业家和工程师，Asif 致力于利用人工智能的潜力造福社会。他最近的创举是推出人工智能媒体平台 Marktechpost，该平台以对机器学习和深度学习新闻的深入报道而闻名，内容既技术性强又易于广泛受众理解。该平台每月浏览量超过 200 万次，显示了其在受众中的受欢迎程度。"],"translationStatus":"translated","bodyOrigin":"source-page","editorial":{"summary":"Aioga 编辑摘要：Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 Aioga 将其归入「产品更新」方向，重点关注它对真实使用和行业竞争的影响。","background":"背景分析：产品与工具类动态的价值取决于它是否解决明确场景、能否进入工作流，以及交付、价格和数据安全是否可接受。","viewpoint":"Aioga 判断：这条动态更适合作为行业观察信号，当前信息足以建立线索，但不足以推导长期结论。","implications":"影响分析：对相关团队而言，短期应先核对来源、可用范围和实际成本，再判断是否值得接入或跟进。","nextStep":"后续观察：继续观察产品是否开放使用、用户反馈、定价、集成能力和后续版本更新。","evidenceRefs":["title","summary","articleBody"],"confidence":"medium","status":"published","aiGenerated":false,"autoApproved":true,"generatedBy":"rule-safe-fallback","generatedAt":"2026-07-28T06:29:10.514Z","sourceHash":"ac006b9187f0629a","validation":{"passed":true,"mode":"rule-safe-fallback","checks":["schema","length","source-attribution","no-html"]}},"tags":["产品更新","MarkTechPost（RSS）"],"translations":{"zh-CN":{"title":"Datalab Marker v2 发布：三模式管道在 olmOCR-bench 上达 76.0，速度超 MinerU 5 倍以上","summary":"Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 倍以上，同时在准确率和速度上均优于 Docling。","category":"产品更新","source":"marktechpost.com","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 发布：三模式管道在 olmOCR-bench 上达 76.0，速度超 MinerU 5 倍以上 - Aioga AI资讯","description":"Datalab 将 Marker 重写为三模式管道，v2 版本在 olmOCR-bench 上达到 76.0 分，单张 B200 上每秒处理 2.9 页--速度是 MinerU 管道后端的 5 倍以上，同时在准确率和速度上均优于 Docling。","url":"https://www.aioga.com/news/cmrzwd6wi01nmrop1ltzp5nza/"},"en":{"title":"Datalab Marker v2 Released: Three-mode Pipeline Achieves 76.0 on olmOCR-bench, Speed Over 5 Times Faster Than MinerU","summary":"Datalab has rewritten Marker as a three-mode pipeline. The v2 version scored 76.0 on olmOCR-bench, processing 2.9 pages per second on a single B200—more than 5 times faster than the MinerU pipeline backend, while outperforming Docling in both accuracy and speed.","category":"Products","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 Released: Three-mode Pipeline Achieves 76.0 on olmOCR-bench, Speed Over 5 Times Faster Than MinerU - Aioga AI News","description":"Datalab has rewritten Marker as a three-mode pipeline. The v2 version scored 76.0 on olmOCR-bench, processing 2.9 pages per second on a single B200—more than 5 times faster than th...","url":"https://www.aioga.com/en/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:42:19.853Z"},"ja":{"title":"Datalab Marker v2 発表：三モードパイプラインが olmOCR-bench で 76.0 を達成、速度は MinerU の5倍以上","summary":"Datalab は Marker を三モードパイプラインに書き換え、v2 バージョンは olmOCR-bench で 76.0 点を達成し、単一の B200 で毎秒 2.9 ページを処理します--速度は MinerU パイプラインのバックエンドの 5 倍以上であり、同時に精度と速度の両方で Docling より優れています。","category":"製品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 発表：三モードパイプラインが olmOCR-bench で 76.0 を達成、速度は MinerU の5倍以上 - Aioga AIニュース","description":"Datalab は Marker を三モードパイプラインに書き換え、v2 バージョンは olmOCR-bench で 76.0 点を達成し、単一の B200 で毎秒 2.9 ページを処理します--速度は MinerU パイプラインのバックエンドの 5 倍以上であり、同時に精度と速度の両方で Docling より優れています。","url":"https://www.aioga.com/ja/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:42:40.260Z"},"ko":{"title":"Datalab Marker v2 출시: 세 가지 모드 파이프라인이 olmOCR-bench에서 76.0 달성, 속도는 MinerU보다 5배 이상 빠름","summary":"Datalab는 Marker를 삼중 모드 파이프라인으로 재작성했으며, v2 버전은 olmOCR-bench에서 76.0점을 기록했고, 단일 B200에서 초당 2.9페이지를 처리할 수 있습니다—속도는 MinerU 파이프라인 백엔드보다 5배 이상 빠르며, 정확도와 속도 모두에서 Docling보다 우수합니다.","category":"제품 업데이트","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 출시: 세 가지 모드 파이프라인이 olmOCR-bench에서 76.0 달성, 속도는 MinerU보다 5배 이상 빠름 - Aioga AI 뉴스","description":"Datalab는 Marker를 삼중 모드 파이프라인으로 재작성했으며, v2 버전은 olmOCR-bench에서 76.0점을 기록했고, 단일 B200에서 초당 2.9페이지를 처리할 수 있습니다—속도는 MinerU 파이프라인 백엔드보다 5배 이상 빠르며, 정확도와 속도 모두에서 Docling보다 우수합니다.","url":"https://www.aioga.com/ko/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:43:23.128Z"},"es":{"title":"Datalab Marker v2 lanzado: el pipeline de tres modos alcanza 76.0 en olmOCR-bench, con una velocidad más de 5 veces superior a MinerU","summary":"Datalab reescribió Marker como una canalización de tres modos, la versión v2 alcanzó 76.0 puntos en olmOCR-bench, procesando 2.9 páginas por segundo en una sola B200, lo que es más de 5 veces la velocidad del backend de la canalización MinerU, y al mismo tiempo supera a Docling tanto en precisión como en velocidad.","category":"Productos","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 lanzado: el pipeline de tres modos alcanza 76.0 en olmOCR-bench, con una velocidad más de 5 veces superior a MinerU - Aioga Noticias de IA","description":"Datalab reescribió Marker como una canalización de tres modos, la versión v2 alcanzó 76.0 puntos en olmOCR-bench, procesando 2.9 páginas por segundo en una sola B200, lo que es más...","url":"https://www.aioga.com/es/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:43:19.421Z"},"fr":{"title":"Datalab Marker v2 publié : le pipeline à trois modes atteint 76,0 sur olmOCR-bench, avec une vitesse plus de 5 fois supérieure à MinerU","summary":"Datalab a réécrit Marker en tant que pipeline à trois modes, la version v2 atteint 76,0 points sur olmOCR-bench, traite 2,9 pages par seconde sur un seul B200 — une vitesse plus de 5 fois supérieure au backend du pipeline MinerU, tout en surpassant Docling à la fois en précision et en vitesse.","category":"Produits","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 publié : le pipeline à trois modes atteint 76,0 sur olmOCR-bench, avec une vitesse plus de 5 fois supérieure à MinerU - Aioga Actualités IA","description":"Datalab a réécrit Marker en tant que pipeline à trois modes, la version v2 atteint 76,0 points sur olmOCR-bench, traite 2,9 pages par seconde sur un seul B200 — une vitesse plus de...","url":"https://www.aioga.com/fr/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:44:06.809Z"},"de":{"title":"Datalab Marker v2 veröffentlicht: Drei-Modi-Pipeline erreicht auf olmOCR-bench 76,0, Geschwindigkeit mehr als 5-mal schneller als MinerU","summary":"Datalab hat Marker in eine Drei-Modus-Pipeline umgeschrieben, die Version v2 erreicht auf olmOCR-bench 76,0 Punkte, verarbeitet auf einem einzelnen B200 2,9 Seiten pro Sekunde – die Geschwindigkeit ist mehr als fünf Mal höher als bei der MinerU-Pipeline, und gleichzeitig übertrifft sie Docling sowohl in Genauigkeit als auch in Geschwindigkeit.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 veröffentlicht: Drei-Modi-Pipeline erreicht auf olmOCR-bench 76,0, Geschwindigkeit mehr als 5-mal schneller als MinerU - Aioga KI-News","description":"Datalab hat Marker in eine Drei-Modus-Pipeline umgeschrieben, die Version v2 erreicht auf olmOCR-bench 76,0 Punkte, verarbeitet auf einem einzelnen B200 2,9 Seiten pro Sekunde – di...","url":"https://www.aioga.com/de/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:44:09.310Z"},"pt-BR":{"title":"Datalab Marker v2 lançado: Pipeline de três modos atinge 76,0 no olmOCR-bench, velocidade mais de 5 vezes maior que MinerU","summary":"Datalab reescreveu o Marker como um pipeline de três modos. A versão v2 alcançou 76,0 pontos no olmOCR-bench, processando 2,9 páginas por segundo em uma única B200 -- velocidade mais de 5 vezes maior que a do back-end do pipeline MinerU, ao mesmo tempo em que supera o Docling tanto em precisão quanto em velocidade.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 lançado: Pipeline de três modos atinge 76,0 no olmOCR-bench, velocidade mais de 5 vezes maior que MinerU - Aioga Notícias de IA","description":"Datalab reescreveu o Marker como um pipeline de três modos. A versão v2 alcançou 76,0 pontos no olmOCR-bench, processando 2,9 páginas por segundo em uma única B200 -- velocidade ma...","url":"https://www.aioga.com/pt-BR/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:44:49.668Z"},"ru":{"title":"Datalab Marker v2 выпущен: трёхрежимный конвейер достигает 76,0 на olmOCR-bench, скорость более чем в 5 раз выше, чем у MinerU","summary":"Datalab переписал Marker как трехрежимный конвейер, версия v2 набрала 76,0 балла на olmOCR-bench, обрабатывая 2,9 страниц в секунду на одном B200 — скорость более чем в 5 раз выше, чем у бэкэнда конвейера MinerU, при этом превосходя Docling как по точности, так и по скорости.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 выпущен: трёхрежимный конвейер достигает 76,0 на olmOCR-bench, скорость более чем в 5 раз выше, чем у MinerU - Aioga Новости ИИ","description":"Datalab переписал Marker как трехрежимный конвейер, версия v2 набрала 76,0 балла на olmOCR-bench, обрабатывая 2,9 страниц в секунду на одном B200 — скорость более чем в 5 раз выше,...","url":"https://www.aioga.com/ru/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:44:58.061Z"},"ar":{"title":"Datalab Marker v2 الإصدار: خط أنابيب ثلاثي الوضعيات يصل إلى 76.0 على olmOCR-bench، والسرعة تفوق MinerU بأكثر من 5 مرات","summary":"قام Datalab بإعادة كتابة Marker إلى أنبوب ثلاثي الأنماط، ونسخة v2 وصلت إلى 76.0 نقطة على olmOCR-bench، ومعالجتها 2.9 صفحة في الثانية على B200 واحدة - السرعة أكثر من 5 أضعاف خلفية أنبوب MinerU، وفي نفس الوقت تتفوق على Docling من حيث الدقة والسرعة.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 الإصدار: خط أنابيب ثلاثي الوضعيات يصل إلى 76.0 على olmOCR-bench، والسرعة تفوق MinerU بأكثر من 5 مرات - Aioga أخبار الذكاء الاصطناعي","description":"قام Datalab بإعادة كتابة Marker إلى أنبوب ثلاثي الأنماط، ونسخة v2 وصلت إلى 76.0 نقطة على olmOCR-bench، ومعالجتها 2.9 صفحة في الثانية على B200 واحدة - السرعة أكثر من 5 أضعاف خلفية أ...","url":"https://www.aioga.com/ar/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:45:45.049Z"},"hi":{"title":"Datalab Marker v2 जारी: तीन मोड पाइपलाइन olmOCR-bench पर 76.0 तक पहुंची, गति MinerU से 5 गुना से अधिक तेज़","summary":"Datalab ने Marker को तीन-मोड़ पाइपलाइन के रूप में फिर से लिखा, v2 संस्करण ने olmOCR-bench पर 76.0 अंक हासिल किए, एकल B200 पर प्रति सेकंड 2.9 पृष्ठ संसाधित किए -- गति MinerU पाइपलाइन बैकएंड की तुलना में 5 गुना से अधिक है, साथ ही सटीकता और गति दोनों में Docling से बेहतर है।","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 जारी: तीन मोड पाइपलाइन olmOCR-bench पर 76.0 तक पहुंची, गति MinerU से 5 गुना से अधिक तेज़ - Aioga AI समाचार","description":"Datalab ने Marker को तीन-मोड़ पाइपलाइन के रूप में फिर से लिखा, v2 संस्करण ने olmOCR-bench पर 76.0 अंक हासिल किए, एकल B200 पर प्रति सेकंड 2.9 पृष्ठ संसाधित किए -- गति MinerU पाइपलाइ...","url":"https://www.aioga.com/hi/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:45:42.262Z"},"it":{"title":"Datalab Marker v2 pubblicato: la pipeline a tre modalità raggiunge 76,0 su olmOCR-bench, con una velocità superiore di oltre 5 volte rispetto a MinerU","summary":"Datalab ha riscritto Marker come una pipeline a tre modalità, la versione v2 ha raggiunto 76,0 punti su olmOCR-bench, elaborando 2,9 pagine al secondo su un singolo B200 -- una velocità più di 5 volte superiore al backend della pipeline MinerU, superando al contempo Docling sia in precisione che in velocità.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 pubblicato: la pipeline a tre modalità raggiunge 76,0 su olmOCR-bench, con una velocità superiore di oltre 5 volte rispetto a MinerU - Aioga Notizie IA","description":"Datalab ha riscritto Marker come una pipeline a tre modalità, la versione v2 ha raggiunto 76,0 punti su olmOCR-bench, elaborando 2,9 pagine al secondo su un singolo B200 -- una vel...","url":"https://www.aioga.com/it/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:46:27.374Z"},"nl":{"title":"Datalab Marker v2 uitgebracht: driefasige pijplijn bereikt 76,0 op olmOCR-bench, snelheid meer dan 5 keer sneller dan MinerU","summary":"Datalab heeft Marker herschreven als een driefasenpijplijn, versie v2 behaalde 76,0 punten op olmOCR-bench, verwerkt 2,9 pagina's per seconde op een enkele B200 -- de snelheid is meer dan vijf keer die van de MinerU-pijplijnback-end, en tegelijkertijd is hij zowel in nauwkeurigheid als snelheid beter dan Docling.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 uitgebracht: driefasige pijplijn bereikt 76,0 op olmOCR-bench, snelheid meer dan 5 keer sneller dan MinerU - Aioga AI-nieuws","description":"Datalab heeft Marker herschreven als een driefasenpijplijn, versie v2 behaalde 76,0 punten op olmOCR-bench, verwerkt 2,9 pagina's per seconde op een enkele B200 -- de snelheid is m...","url":"https://www.aioga.com/nl/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:46:29.005Z"},"tr":{"title":"Datalab Marker v2 Yayınlandı: Üç modlu boru hattı olmOCR-bench üzerinde %76.0'e ulaştı, hızı MinerU'dan 5 kat daha fazla","summary":"Datalab, Marker'ı üç modlu bir boru hattı olarak yeniden yazdı. v2 sürümü olmOCR-bench üzerinde 76,0 puan elde etti, tek B200 kartında saniyede 2,9 sayfa işledi – bu hız, MinerU boru hattı arka ucundan 5 kat daha hızlı, aynı zamanda doğruluk ve hız açısından Docling'den de üstün.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 Yayınlandı: Üç modlu boru hattı olmOCR-bench üzerinde %76.0'e ulaştı, hızı MinerU'dan 5 kat daha fazla - Aioga AI Haberleri","description":"Datalab, Marker'ı üç modlu bir boru hattı olarak yeniden yazdı. v2 sürümü olmOCR-bench üzerinde 76,0 puan elde etti, tek B200 kartında saniyede 2,9 sayfa işledi – bu hız, MinerU bo...","url":"https://www.aioga.com/tr/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:47:09.629Z"},"vi":{"title":"Datalab Marker v2 ra mắt: Ống ba chế độ đạt 76,0 trên olmOCR-bench, tốc độ nhanh hơn MinerU hơn 5 lần","summary":"Datalab đã viết lại Marker thành pipeline ba chế độ, phiên bản v2 đạt 76,0 điểm trên olmOCR-bench, xử lý 2,9 trang mỗi giây trên B200 đơn lẻ -- tốc độ nhanh hơn hơn 5 lần so với backend pipeline của MinerU, đồng thời vượt trội hơn Docling cả về độ chính xác lẫn tốc độ.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 ra mắt: Ống ba chế độ đạt 76,0 trên olmOCR-bench, tốc độ nhanh hơn MinerU hơn 5 lần - Tin tức AI Aioga","description":"Datalab đã viết lại Marker thành pipeline ba chế độ, phiên bản v2 đạt 76,0 điểm trên olmOCR-bench, xử lý 2,9 trang mỗi giây trên B200 đơn lẻ -- tốc độ nhanh hơn hơn 5 lần so với ba...","url":"https://www.aioga.com/vi/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:47:09.203Z"},"id":{"title":"Datalab Marker v2 Dirilis: Pipeline Tiga Mode Mencapai 76,0 di olmOCR-bench, Kecepatan Lebih Dari 5 Kali MinerU","summary":"Datalab menulis ulang Marker menjadi saluran tiga mode, versi v2 mencapai skor 76,0 di olmOCR-bench, memproses 2,9 halaman per detik pada B200 tunggal -- kecepatannya lebih dari 5 kali lipat dari backend saluran MinerU, sambil unggul dalam hal akurasi dan kecepatan dibandingkan Docling.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 Dirilis: Pipeline Tiga Mode Mencapai 76,0 di olmOCR-bench, Kecepatan Lebih Dari 5 Kali MinerU - Berita AI Aioga","description":"Datalab menulis ulang Marker menjadi saluran tiga mode, versi v2 mencapai skor 76,0 di olmOCR-bench, memproses 2,9 halaman per detik pada B200 tunggal -- kecepatannya lebih dari 5...","url":"https://www.aioga.com/id/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:47:50.629Z"},"th":{"title":"Datalab Marker v2 เปิดตัว: ท่อสามโหมดบน olmOCR-bench ทำได้ 76.0 ความเร็วสูงกว่า MinerU ถึงกว่า 5 เท่า","summary":"Datalab ได้เขียน Marker ใหม่เป็นท่อแบบสามโหมด รุ่น v2 ทำคะแนนได้ 76.0 บน olmOCR-bench โดยประมวลผลได้ 2.9 หน้าในหนึ่งวินาทีบน B200 หนึ่งเครื่อง—ความเร็วมากกว่าหลังท่อ MinerU กว่า 5 เท่า ในขณะเดียวกันทั้งความแม่นยำและความเร็วก็เหนือกว่า Docling","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 เปิดตัว: ท่อสามโหมดบน olmOCR-bench ทำได้ 76.0 ความเร็วสูงกว่า MinerU ถึงกว่า 5 เท่า - ข่าว AI Aioga","description":"Datalab ได้เขียน Marker ใหม่เป็นท่อแบบสามโหมด รุ่น v2 ทำคะแนนได้ 76.0 บน olmOCR-bench โดยประมวลผลได้ 2.9 หน้าในหนึ่งวินาทีบน B200 หนึ่งเครื่อง—ความเร็วมากกว่าหลังท่อ MinerU กว่า 5...","url":"https://www.aioga.com/th/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:48:02.991Z"},"pl":{"title":"Datalab Marker v2 wydany: Potrójny tryb rurociągu osiągnął 76,0 na olmOCR-bench, prędkość ponad 5 razy większa niż MinerU","summary":"Datalab przepisał Marker na potrójny tryb pipeline, wersja v2 osiągnęła 76,0 punktów na olmOCR-bench, przetwarzając 2,9 strony na sekundę na pojedynczym B200 – prędkość jest ponad 5 razy większa niż back-end pipeline MinerU, jednocześnie przewyższając Docling zarówno pod względem dokładności, jak i prędkości.","category":"产品更新","source":"MarkTechPost（RSS）","aggregationSource":"MarkTechPost（RSS）","pageTitle":"Datalab Marker v2 wydany: Potrójny tryb rurociągu osiągnął 76,0 na olmOCR-bench, prędkość ponad 5 razy większa niż MinerU - Aioga Wiadomości AI","description":"Datalab przepisał Marker na potrójny tryb pipeline, wersja v2 osiągnęła 76,0 punktów na olmOCR-bench, przetwarzając 2,9 strony na sekundę na pojedynczym B200 – prędkość jest ponad...","url":"https://www.aioga.com/pl/news/cmrzwd6wi01nmrop1ltzp5nza/","contentTranslated":true,"sourceHash":"7eee76ff7bb33847","translatedAt":"2026-07-26T16:48:46.535Z"}}}}