[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-groundingpi-4b-visual-grounding-foundation-model":3,"topics-all":38,"news-related-5d4acf64-67d8-404b-9d3f-cdf7761fb8ac":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},"5d4acf64-67d8-404b-9d3f-cdf7761fb8ac","小鹏开源 GroundingPI:4B 视觉 grounding 模型自评超 GPT-6 Astra","XPENG AI 开源 4B 视觉 grounding 模型 GroundingPI。官方在 34 项基准平均 73.68%,自评超 GPT-6 Astra;作为机器人与自动驾驶视觉骨干,分布外测试最高相对提升 24.8%。基于 Qwen3-4B 主干,单卡 vLLM 可部署,Apache 2.0 开源。","机器人与自动驾驶圈有个长期痛点:VLA(vision-language-action)和 world-action 模型的感知,大多直接搬通用视觉语言或视频生成 backbone,可这些 backbone 在杂乱场景、小目标、闭环控制要的速度上频频失手。XPENG AI(小鹏)团队 9 月 30 日放出的 GroundingPI,就是冲着这个缺口来的:一个 4B 参数的视觉 grounding 基础模型,论文当天上 arXiv,权重、源码、部署与批量标注指南随后十天全部到位。\n\n## 4B 参数,34 项基准自评登顶\n\nGroundingPI 做的事很纯粹:给定一张图和一句指令,输出点或框,坐标在共享词表里量化成 0-999 网格上的整数 token(GAM spatial-token 协议)。训练分三段——多模态与空间预训练、监督微调、GRPO 强化学习,监督信号来自公开数据集加自建数据引擎。官方在 34 项 grounding 基准、44 个 baseline、11 项感知能力上做了评测:平均 73.68%,论文自评超过更大的 GPT-6 Astra(71.54%),并称建立了新的 state of the art。一作还在 Hugging Face 论文页评论区补充:用过 Grounding DINO 的,可以直接换它试试。\n\n## 架构全是开源件,交付按工程标准来\n\n拆开看,GroundingPI 的三个组件都是现成货:MoonViT-V2(出自 Kimi K3)视觉编码器、可学习多模态 projector、Qwen3-4B 语言主干。思路不是发明新架构,而是把「感知」当成一等公民专门预训练。交付侧相当工程化:vLLM Transformers 后端部署、OpenAI 兼容 API、BF16 精度、16384 上下文、TP=1 单卡即可跑,官方标注的测试硬件覆盖 NVIDIA B300\u002FB200\u002FH200\u002FH800 和 PPU。任务面覆盖检测、指代框、点定位、OCR、文档版式、GUI grounding。许可为 Apache 2.0,Kimi 衍生部分另守 Kimi K3 License。\n\n## 下游才是真正的卖点\n\n单看 grounding 分数只是开胃菜,论文的重头在下游迁移:RoboTwin 2.0 上,它作为视觉 backbone 在全部 4 个分布外设定里超过所有参评主流 backbone,最高相对提升 24.8%;RoboCasa-GR1 上,用 50% 演示数据训练就能压过用 75% 数据的 baseline——数据效率的信号比绝对分数更值钱;nuScenes 自动驾驶开环测试平均 L2 误差 0.296 米。\n\n## 自评成绩单,冷水和信号各一半\n\n必须说清楚:73.68% 超 GPT-6 Astra 这类声明,目前全部来自官方自评,论文与 GitHub README 两份官方工件互相印证,尚无第三方独立复现,读的时候心里打个折。不过方向本身值得认真对待——通用 VLM 当感知器是「什么都行、什么都不精」,把感知单独拿出来专门预训练,本质是给具身系统配一个 System-1 基础模型,和世界模型派「快慢双系统」的叙事同向。另一个更有信号量的细节:小鹏今年已经连续放出 TuringViT 视觉基座、X-AuT 车载语音剪枝、再到这次的 GroundingPI,车企 AI 团队在做「基座化」输出,这股趋势比任何单个模型的分数都更值得关注。对具身和自动驾驶团队来说,现在多了一个可上手的开源感知基座;对通用 VLM 团队来说,「感知专门化」这条岔路值得盯住。\n\n参考:论文 arXiv:2609.39601;代码 github.com\u002Fgroundingpi\u002FGroundingPI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.39601","7437aeb9-930c-4866-a2e9-48003c1a792b",[11,15,18,21],{"id":12,"name":13,"slug":13,"description":14,"color":14},"01598627-1ea6-4b27-a5d8-874971571a71","llm",null,{"id":16,"name":17,"slug":17,"description":14,"color":14},"7e89b5cc-57db-4f37-bc6d-28919a73931c","model-release",{"id":19,"name":20,"slug":20,"description":14,"color":14},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":22,"name":23,"slug":23,"description":14,"color":14},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[25],{"id":26,"lang":27,"title":28,"summary":29,"content":30},"6fdfafe4-a730-47fe-9e32-52fe893b5de3","en","XPENG Open-Sources GroundingPI: 4B Grounding Model, 73.68% Avg","XPENG open-sources GroundingPI, a 4B visual grounding model averaging 73.68% over 34 benchmarks, self-rated above GPT-6 Astra.","Robotics and autonomous driving teams keep running into the same wall: vision-language-action (VLA) and world-action models borrow their perception from general-purpose vision-language or video-generation backbones, and those backbones fail in clutter, on tiny objects, and at the speed closed-loop control demands. GroundingPI, released by XPENG AI on September 30, targets exactly that gap: a 4B-parameter visual grounding foundation model, with the paper on arXiv the same day, weights opened October 1, and source code plus deployment and batch-annotation guides following on October 3.\n\n## A 4B model topping 34 benchmarks on self-report\n\nThe task is deliberately narrow: given an image and an instruction, output points or boxes, with coordinates quantized as integer tokens on a 0-999 grid (the GAM spatial-token protocol). Training proceeds in three stages — multimodal and spatial pretraining, supervised fine-tuning, and GRPO reinforcement learning — with supervision from public datasets plus dedicated data engines. The official evaluation spans 34 grounding benchmarks, 44 baselines, and 11 perceptual capabilities: GroundingPI averages 73.68%, which the paper self-reports above the larger GPT-6 Astra at 71.54%, claiming a new state of the art. The first author added a positioning note in the Hugging Face paper comments: teams using Grounding DINO can try this as a drop-in upgrade — 4B parameters, with a complete inference stack.\n\n## Architecture from off-the-shelf parts, engineering-grade delivery\n\nThe three components are all existing open-source pieces: a MoonViT-V2 visual encoder (from Kimi K3), a learnable multimodal projector, and a Qwen3-4B language backbone. The idea is not a new architecture but treating perception as a first-class citizen for dedicated pretraining. Delivery is unusually engineering-grade: vLLM Transformers-backend serving, an OpenAI-compatible API, BF16 precision, a 16,384-token context, and TP=1 on a single card. Officially tested accelerators cover NVIDIA B300, B200, H200, H800, and PPU. Task coverage spans detection, referring boxes, point localization, OCR, document layout, GUI grounding, and visual prompting. The license is Apache 2.0, with Kimi-derived material governed by the Kimi K3 License.\n\n## Downstream transfer is the real story\n\nGrounding scores are the appetizer; the main course is downstream transfer. On RoboTwin 2.0, GroundingPI as the visual backbone beats every mainstream backbone evaluated across all four out-of-distribution settings, by up to 24.8% relative to the strongest one. On RoboCasa-GR1, training with 50% of the demonstrations outperforms baselines trained with 75% — a data-efficiency signal worth more than any absolute score. On nuScenes, used as the visual backbone for autonomous driving, it reaches an average open-loop L2 error of 0.296 meters.\n\n## A self-graded report card: half cold water, half signal\n\nTo be clear: the 73.68%-beats-GPT-6-Astra claim currently comes entirely from the official self-evaluation, cross-checked between two official artifacts (the paper and the GitHub README) but with no independent third-party replication yet — discount accordingly. The direction, however, deserves attention. Using general VLMs as perceptual engines means decent-at-everything, sharp-at-nothing; carving perception out for dedicated pretraining is essentially building a System-1 foundation model for embodied stacks, aligned with the fast\u002Fslow dual-system narrative of the world-model camp. One more detail carries more signal than any single score: XPENG has this year shipped TuringViT, the X-AuT in-car speech pruning work, and now GroundingPI — automotive AI teams are shipping foundation-model artifacts, and that trend outlasts any leaderboard. For embodied and autonomous-driving teams, there is now a 4B, vLLM-deployable, permissively licensed perception backbone to try; for general VLM teams, the specialized-perception detour is worth watching before it overtakes.\n\nReferences: paper at arXiv:2609.39601 (https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.39601); code and weights at https:\u002F\u002Fgithub.com\u002Fgroundingpi\u002FGroundingPI","groundingpi-4b-visual-grounding-foundation-model","2026-10-13T17:10:26Z","2026-10-13T17:10:40.087661Z","2026-10-13T17:10:40.087675Z",true,"agent",90,[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},"cd49f913-cde7-4cf3-8d93-24508653180e","腾讯混元开源AuK:1.5B语音模型统一生成与编辑,4步推理快4.5倍","tencent-hunyuan-auk-speech-editing","2026-09-09T09:12:00+00:00",{"id":65,"title":66,"news_slug":67,"published_at":68},"40095b51-97b0-4fd4-9b1d-f636c970572e","阿里 Qwen 团队发布 Qwen3.8-Max:2.4 万亿参数 MoE 模型首度开放权重","qwen3-8-max-2-4t-moe-open-weights","2026-08-07T02:00:00+00:00",{"id":70,"title":71,"news_slug":72,"published_at":73},"804b44fb-66c6-4355-83a4-b3a03a776d2a","Inkling-Small 开放权重：12B 激活参数换来更高 Agent 效率，也暴露事实性短板","inkling-small-multimodal-moe-efficiency","2026-08-05T16:32:13+00:00",{"id":75,"title":76,"news_slug":77,"published_at":78},"a151db0c-d832-4df2-ac03-2d4e58b26e99","Kimi K3 跑通 MiniTriton:Moonshot 让 LLM 第一次从零编译出自己的 GPU 编译器","kimi-k3-minitriton-gpu-compiler","2026-07-26T14:00:00+00:00",{"id":80,"title":81,"news_slug":82,"published_at":83},"dfc3dec4-2211-4c7e-b6ff-9e0d9a479ec4","微软与 Mistral 签下数十亿美元协议:Vera Rubin GPU 上的「欧洲主权云」开始落地","microsoft-mistral-vera-rubin-sovereign","2026-07-22T02:00:00+00:00",{"id":85,"title":86,"news_slug":87,"published_at":88},"cf01282f-8a64-49a8-a608-9b806ccfbea3","Mira Murati 实验室 Inkling 开源：975B MoE 不卷\"最强\"，押注\"可定制\"","thinking-machines-inkling","2026-07-15T22:00:00+00:00"]