[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-omnidr-bench-video-deep-research":3,"topics-all":38,"news-related-5be712c9-ff84-4e5a-aefa-673f880c8ed1":57},{"id":4,"title":5,"summary":6,"content":7,"original_url":8,"source_id":9,"tags":10,"translations":24,"news_slug":31,"published_at":32,"created_at":33,"modified_at":34,"is_published":35,"publish_type":36,"image_url":14,"view_count":37},"5be712c9-ff84-4e5a-aefa-673f880c8ed1","视频深度研究报告基准:AI排版强内容弱,最高63.3分","北大与阿里Qwen团队发布OmniDR-Bench:以一条视频为研究锚点,让智能体检索外部证据、生成带引用的交互式报告。50个任务、6414条细则、18个配置实测,最高63.3分;排版均值81.8,内容45.4,引用47.8,外部证据使用率仅8.7%——报告好看,研究不足。","深度研究智能体如今不稀奇:丢一个问题,它自己搜网页、写报告、附引用。但把研究的主角换成一条视频,难度立刻不一样——证据一半在画面和声音里,一半在画面之外,还得把两边对上。10月22日,北京大学与阿里巴巴Token Hub的团队(一作为Qwen团队实习生)在arXiv放出OmniDR-Bench,专门测「以视频为中心的深度研究报告」。结果不算体面:18个模型配置里,点估最高的Claude Opus 5只拿63.3分。\n\n## 任务与评分:6414条原子细则\n\n任务定义直接:给智能体一条源视频加一个用户请求,它要自己提出研究问题,检索网页、图片、视频、文档等外部证据,最后产出一份挂在视频时间轴上、可交互、带引用的HTML报告。论文自己的定位是:此前基准要么测纯文本深研,要么测视频问答短答案,没人评过「围绕一条视频做完整研究报告」这个任务。\n\n评分用他们提出的Omni-ART原子评分树:三个维度、23个共享方向、按任务定制的二元判据,合计6414条细则,覆盖视听定位、多模态证据使用与交互呈现。构造流程走了独立取证、自动检查加两轮人工审核,候选任务从82个砍到60个、再到50个(中英双语、19个真实子类);终版再抽1000条细则人工复核,94%以上判为正确。裁判是GPT-5.6 Sol配浏览器探针做确定性检查,900次运行产出835份可评报告。\n\n## 排版81.8分,内容45.4分\n\n主榜单里,12个模型跑在Claude Code框架上,Codex与OpenClaw各覆盖3个配置。最值得看三行:Claude Opus 5以63.3排第一,但丢了4%的任务;GLM-5.3拿62.0且50个任务全部交付,1.3分的差距落在置信区间[-4.0, 5.6]内,谁先谁后说不准;Qwen3.8-Max单份报告冲到全场最高的91.5,完成率却只有88%,总分被拖到52.9。\n\n更扎心的是维度拆分:在统一报告模板下,呈现维度平均81.8分,内容45.4,引用47.8。智能体把视频本身讲得挺忠实(直接视频事实通过率79.9%,视频忠实度92.4%),但一旦要走出视频做研究就露馅:外部证据使用率8.7%,跨源核实8.8%,多事实组合5.0%,论断支撑只有15.9%。引用行为尤其典型——GPT-5.6 Sol的来源数量判据通过98.0%,正文论断旁的引用支撑却只有23.9%;Gemini-3.8-Flash这两项是93.0%对13.9%。链接堆在文末来源区,正文结论旁边并没有证据。\n\n## 编排与框架:分数是花钱买的\n\n技能消融显示每层能力都有价:只有界面基线28.0分;接入视频理解到31.8;加上网页检索直接跳到45.5(任务成功率从73.3%拉满到100%);再加多模态检索46.6;完整研究编排53.9。完整编排比全部原子技能再涨7.3分(15个匹配任务里12个提升,引用维度+20.7),代价是中位步数从103涨到167、耗时从0.7小时到1.6小时、token从230万涨到470万——质量换成本,明码标价。\n\n执行框架的影响同样不小:同一个GPT-5.6 Sol,Claude Code下54.9,Codex下53.9,OpenClaw只剩41.7;换成Qwen3.7-Max反而是Codex最好(45.3对39.9)。两框架在三个匹配模型上均值同为48.4,但Codex完成率99.3%,Claude Code为96.0%。框架没有绝对赢家,模型和框架要配着选。\n\n## 所以呢\n\n这份基准给「深研产品热」泼的冷水很具体:模板一给,排版人人高分;真正的研究行为——找外部证据、交叉核实、组合多事实——目前几乎没有配置做得像样。对选型的人:看内容分和引用分,别被演示页的渲染效果骗了。对做产品的人:GLM-5.3那种100%交付加62分的稳定性,线上可能比63.3更值钱。基准、代码与运行轨迹已开源,详见[arXiv:2610.31158](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.31158)与[GitHub仓库](https:\u002F\u002Fgithub.com\u002Fmarinero4972\u002FOmniDR-Bench)。","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.31158","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"6ad31a14-c0da-42df-81fd-564281f768db","agentic-ai",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"120fa59a-ff6f-4537-9bf5-f818df636a0e","benchmark",{"id":19,"name":20,"slug":20,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",{"id":22,"name":23,"slug":23,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"98c124a6-fa04-49e5-b8db-c911021e1120","en","OmniDR-Bench: Video Deep Research Agents Peak at 63.3\u002F100","Alibaba Qwen's OmniDR-Bench: video deep-research agents hit 63.3\u002F100 in 18 configs; presentation 81.8 vs content 45.4, external evidence 8.7%.","Deep research agents are no longer novel: give one a question and it searches the web, writes a report, and adds citations. Swap the subject for a single video, though, and the task gets much harder — half the evidence sits inside the footage and its audio, half lies outside it, and the two halves must be reconciled. On October 22, a team from Peking University and Alibaba's Token Hub (first author an intern with the Qwen team) released OmniDR-Bench on arXiv, built to test exactly this: video-centered deep research reports. The outcome is not flattering. Across 18 model configurations, the highest point estimate — Claude Opus 5 — reaches only 63.3.\n\n## The task and the rubric: 6,414 atomic criteria\n\nThe task is direct: an agent receives a source video and a user request, must derive its own research questions, retrieve external evidence from webpages, images, videos and documents, and produce an interactive, cited HTML report anchored to the video timeline. In the authors' framing, existing benchmarks cover either text-only deep research or video question answering with short answers; none evaluated producing a full research report organized around one given video.\n\nScoring uses their Omni-ART atomic rubric trees: three dimensions, 23 shared directions, and task-specific binary criteria — 6,414 criteria in total — covering audiovisual grounding, multimodal evidence use, and interactive presentation. Construction ran through independent evidence collection, automated checks, and two rounds of human review; the candidate pool shrank from 82 tasks to 60 and then to the final 50 (bilingual Chinese-English, 19 real-world subcategories). A final audit sampled 1,000 criteria, with more than 94% judged correct. GPT-5.6 Sol serves as judge, paired with deterministic browser probes; 900 runs yielded 835 evaluable reports.\n\n## Presentation 81.8, content 45.4\n\nTwelve models run on the Claude Code harness, with Codex and OpenClaw each covering three configurations. Three rows stand out. Claude Opus 5 tops the board at 63.3 but misses 4% of tasks. GLM-5.3 scores 62.0 while delivering all 50 tasks; the 1.3-point gap falls inside a confidence interval of [-4.0, 5.6], so the ranking is uncertain. Qwen3.8-Max posts the highest single report at 91.5, yet an 88% completion rate drags its overall score to 52.9.\n\nThe dimension breakdown stings more: under a shared report template, presentation averages 81.8, content 45.4, citation 47.8. Agents retell the video itself faithfully (direct-video-fact pass rate 79.9%, video faithfulness 92.4%), but step outside the video and they fall apart: external evidence use 8.7%, cross-source verification 8.8%, multi-fact composition 5.0%, claim support just 15.9%. The citation pattern is telling — GPT-5.6 Sol passes 98.0% of source-count criteria but only 23.9% of claim-support criteria; for Gemini-3.8-Flash the pair is 93.0% versus 13.9%. Links pile up in a final Sources section while the claims themselves sit unsupported.\n\n## Orchestration and harnesses: quality bought with money\n\nThe skill ablation prices every layer: interface-only baseline 28.0; adding video understanding 31.8; adding web research jumps to 45.5 (task success climbs from 73.3% to 100%); multimodal retrieval 46.6; full research orchestration 53.9. Full orchestration adds 7.3 points over all atomic skills (12 of 15 matched tasks improve, citation up 20.7), at the cost of median steps rising from 103 to 167, time from 0.7 to 1.6 hours, and tokens from 2.3M to 4.7M — a quality-cost trade, explicitly priced.\n\nHarness choice matters too. The same GPT-5.6 Sol scores 54.9 on Claude Code, 53.9 on Codex, and only 41.7 on OpenClaw; swap in Qwen3.7-Max and Codex is suddenly best (45.3 versus 39.9). The two harnesses average an identical 48.4 across three matched models, but Codex completes 99.3% of runs against Claude Code's 96.0%. There is no universal winner — model and harness must be chosen together.\n\n## So what\n\nThe cold water this benchmark pours on deep-research hype is specific: hand agents a template and presentation scores soar; the actual research behaviors — finding external evidence, cross-checking sources, composing multiple facts — barely exist in any configuration. For buyers: read the content and citation scores, not the rendered demo. For builders: GLM-5.3's 100% delivery at 62.0 may be worth more in production than 63.3. The benchmark, code and trajectories are open — see [arXiv:2610.31158](https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.31158) and the [GitHub repo](https:\u002F\u002Fgithub.com\u002Fmarinero4972\u002FOmniDR-Bench).","omnidr-bench-video-deep-research","2026-10-25T19:15:23Z","2026-10-25T19:15:28.346978Z","2026-10-25T19:15:28.346986Z",true,"agent",212,[39,48],{"slug":40,"tag_slug":40,"title_zh":41,"title_en":42,"intro_zh":43,"intro_en":44,"id":45,"is_active":35,"created_at":46,"modified_at":47},"ai-for-science","AI for Science 2026：从 UniPert 到 GPT-Rosalind 的硬核进化","AI for Science 2026: from UniPert to GPT-Rosalind","生命科学、化学材料、物理世界模型——AI 正在从\"语言工具\"变成\"实验伙伴\"。本专题收录 AI 在三大科学方向的关键节点：UniPert 统一基因与化学扰动空间、GPT-Rosalind 端到端生命科学推理、达摩院 AI 智能体 28 小时找到 4 种超导新材料、Anthropic Claude Science 把工作台做成标准品。","From language tool to lab partner — AI is reshaping life sciences, chemistry\u002Fmaterials, and physical world models. This topic covers the key milestones: UniPert unifying genetic-chemical perturbation spaces, GPT-Rosalind's end-to-end life-sciences reasoning, DAMO's AI agent discovering 4 superconducting materials in 28 hours, and Anthropic's Claude Science workbench going mainstream.","988a4300-5fab-41c4-b5d8-63711a2dc757","2026-09-10T01:34:15.296649Z","2026-09-10T01:34:15.296663Z",{"slug":49,"tag_slug":49,"title_zh":50,"title_en":51,"intro_zh":52,"intro_en":53,"id":54,"is_active":35,"created_at":55,"modified_at":56},"h3-series","MiniMax H3 系列：从开源权重到 35 倍吞吐","MiniMax H3 Series: from open weights to 35x throughput","MiniMax H3 自 2026 年 8 月开源以来节奏密集：官方把生成、参考与编辑收回一个模型；ComfyUI 当天压进 RTX 3060；摩尔线程 3 小时完成国产 GPU 适配；fal 后训练版把吞吐拉到 35 倍；FastH3 蒸馏再砍推理成本。本专题持续追踪 H3 的发布—开源—蒸馏—部署全链路。","Since MiniMax open-sourced H3 in August 2026 the pace has been relentless: one unified omni-modal model, same-day ComfyUI support down to an RTX 3060, a 3-hour Day-0 port to Moore Threads GPUs, fal's post-trained H3 Max at 35x throughput, and FastH3 distillation cutting inference cost further. This topic tracks the full H3 chain — release, open weights, distillation, deployment.","83ef0daa-3c31-4cb3-86ed-e5ee58654d5f","2026-09-08T07:33:19.942193Z","2026-09-08T07:33:19.942209Z",{"items":58},[59,64,69,74,79,84],{"id":60,"title":61,"news_slug":62,"published_at":63},"fd0e593c-bb2c-44f9-8913-16f8dcef6999","苹果8B智能体:真实iOS上追平Gemini-3.1-Pro","apple-online-rl-ui-agent-live-ios","2026-10-31T11:10:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"205bd813-42eb-4d65-8e8c-5a9af0d8d097","12个模型搜遍真实网页,最高分39%:人类21.1%更低","opends-world-benchmark-deep-search-agents","2026-10-23T11:20:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"4f5c4072-660c-47e7-9d9a-9782eb5591bf","Pistis 报告:IDRL 让蒸馏和 RL 交替上岗","pistis-idrl-interleaved-distillation-rl","2026-09-25T19:11:33+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"2f9ad534-b737-490e-b97e-bf01f2b6d26b","交错生成新基准:闭源旗舰文本赢,图像一致性输给开源","intreal-interleaved-generation-benchmark","2026-11-01T05:27:08+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"6f2c5b92-a390-4b36-9eff-534daf04f4aa","WorkFactory:先修任务再微调,879涨到1565","workfactory-workpilot-agent-training","2026-10-31T23:10:20+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"9109e6b0-bf85-472d-abf6-ba8e31dc17c5","Co-Design 基准:七成准则过,26.4% 完整交付","co-design-bench-agent-delivery-gap","2026-10-30T05:00:00+00:00"]