[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"news-slug-ant-lingbot-vla-2-0":3,"topics-all":36,"news-related-5de8c559-2ab5-44a7-b1d6-97cc8e499b25":55},{"id":4,"title":5,"summary":6,"content":6,"original_url":7,"source_id":8,"tags":9,"translations":23,"news_slug":29,"published_at":30,"created_at":31,"modified_at":32,"is_published":33,"publish_type":34,"image_url":13,"view_count":35},"5de8c559-2ab5-44a7-b1d6-97cc8e499b25","蚂蚁灵波开源 LingBot-VLA 2.0：6 万小时数据 + 17 个品牌,把具身基座卷向跨构型","7 月 8 日,蚂蚁灵波科技宣布升级并开源新一代具身基座模型 LingBot-VLA 2.0,这是继今年 1 月 LingBot-VLA 1.0 之后的全面迭代。\n\n相比 1.0 版本,2.0 最直观的变化在数据规模和构型覆盖。在预训练阶段,模型融入了 6 万小时高质量真实物理数据,覆盖 17 个主流机器人品牌的 20 多种机器人构型。这背后解决的是具身智能的「跨构型泛化」老难题——以往 VLA 模型往往只能针对单一品牌或单一形态调优,换一个机器人就要从头微调。\n\n更重要的是自由度拓展:LingBot-VLA 2.0 新增对头部、腰部、末端执行器乃至移动底盘自由度的支持。这意味着同一个基座模型,既能操控固定臂的工业机械臂,也能驱动带移动底盘的服务机器人,甚至可以处理多自由度协同任务。\n\n蚂蚁灵波这次选择「开源」的时机并不偶然。VLA 类模型已进入「量产试水」阶段,从 Pi 0 到 RDT、NeuroVLA、HY-VLA,各家都在抢工厂与具身方案商的入口。开源一个跨 17 个品牌、能跨构型复用的基座,等于把自己做成「具身时代的 HuggingFace 候选」——通过占据工具链上游,去影响下游机器人的部署选型。\n\nLingBot-VLA 2.0 真正有意思的地方,不是数据量本身,而是它把「多构型 + 多自由度」压进同一个端到端基座模型里——这意味着具身智能正在走一条「粗统一、再细调」的路径,而不是永远靠堆叠专用模型。","https:\u002F\u002F36kr.com\u002Fnewsflashes\u002F3886479015555336","5e4fd3d1-9cb4-44a6-bae5-9ffb449c05c1",[10,14,17,20],{"id":11,"name":12,"slug":12,"description":13,"color":13},"471c51be-e620-49df-bd6c-0b5504f53f00","ant-group",null,{"id":15,"name":16,"slug":16,"description":13,"color":13},"499f4b56-819d-49a3-9609-33e775143b86","multimodal",{"id":18,"name":19,"slug":19,"description":13,"color":13},"b1853a5a-d940-42b7-94f9-0488ee3f2cf7","new-model",{"id":21,"name":22,"slug":22,"description":13,"color":13},"b9bd9039-fcdb-41a8-b85b-fc1587def2b9","open-source",[24],{"id":25,"lang":26,"title":27,"summary":28,"content":13},"d1e39963-ebd2-49a0-bb9f-066e6fe9d7dd","en","LingBot-VLA 2.0: 60K hours across 17 robot brands","On July 8, Ant Group's Lingbo Tech announced the upgrade and open-source release of its new-generation embodied base model LingBot-VLA 2.0, a comprehensive iteration after LingBot-VLA 1.0 in January this year. Compared to version 1.0, the most visible change is in data scale and form-factor coverage. In the pretraining phase, the model incorporates 60K hours of high-quality real physical data, covering 17 mainstream robot brands and 20+ robot form factors. What this addresses is the old embodied-AI problem of \"cross-form-factor generalization\" — VLA models in the past could usually only be tuned for a single brand or single form, requiring from-scratch fine-tuning to switch to another robot. More important is the degrees-of-freedom expansion: LingBot-VLA 2.0 adds support for the degrees of freedom of head, waist, end-effector, and even mobile chassis. This means the same base model can manipulate both fixed-arm industrial robotic arms and service robots with mobile chassis, and can even handle multi-degree-of-freedom coordinated tasks. Ant Lingbo's choice of \"open source\" timing is not accidental. VLA-class models have entered the \"trial mass production\" stage, from Pi 0 to RDT, NeuroVLA, HY-VLA, every player is fighting for the entry point of factories and embodied solution providers. Open-sourcing a cross-17-brand, cross-form-factor reusable base is equivalent to making itself a \"candidate for the embodied-era HuggingFace\" — by occupying the upstream of the toolchain, influencing the deployment choices of downstream robots. What's really interesting about LingBot-VLA 2.0 isn't the data volume itself, but the fact that it compresses \"multi-form-factor + multi-degree-of-freedom\" into the same end-to-end base model — this means embodied intelligence is walking a \"rough unification, then fine-tuning\" path, rather than forever stacking dedicated models.","ant-lingbot-vla-2-0","2026-07-08T06:30:00Z","2026-07-08T06:06:41.385390Z","2026-08-19T02:08:40.142862Z",true,"agent",167,[37,46],{"slug":38,"tag_slug":38,"title_zh":39,"title_en":40,"intro_zh":41,"intro_en":42,"id":43,"is_active":33,"created_at":44,"modified_at":45},"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":47,"tag_slug":47,"title_zh":48,"title_en":49,"intro_zh":50,"intro_en":51,"id":52,"is_active":33,"created_at":53,"modified_at":54},"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":56},[57,62,67,72,77,82],{"id":58,"title":59,"news_slug":60,"published_at":61},"c99751d5-418e-49d5-99d3-e43b84c80ec7","IBM与NASA开源月球基础模型:Lunar Foundation Model","nasa-ibm-lunar-foundation-model-sombench","2026-09-19T09:30:00+00:00",{"id":63,"title":64,"news_slug":65,"published_at":66},"51c13e24-8072-404c-a8d4-75c40cff05ee","Ling-3.0-flash-VL 开源：124B MoE 只激活 5.5B，视觉塞进 Agent 闭环","ling-3-0-flash-vl-open-weights","2026-09-15T13:18:00+00:00",{"id":68,"title":69,"news_slug":70,"published_at":71},"e3c0b314-d7b7-4901-b2b0-08ca5ef08ac7","GigaBrain-0.7开源:37k小时数据+三系统架构,世界模型进VLA决策回路","gigabrain-0-7-embodied-vla-open-source","2026-08-26T23:15:00+00:00",{"id":73,"title":74,"news_slug":75,"published_at":76},"d8c62859-54c8-4069-b776-8e623ca03029","Cohere Transcribe Arabic：2B 开源 ASR 登顶，WER 低 Whisper 11 点","cohere-transcribe-arabic","2026-07-16T04:00:00+00:00",{"id":78,"title":79,"news_slug":80,"published_at":81},"c677680a-a08f-420c-8107-7816827707a2","小米开源 Xiaomi-Robotics-U0：38B 具身生成统一 Tokenizer","xiaomi-robotics-u0","2026-07-14T22:10:00+00:00",{"id":83,"title":84,"news_slug":85,"published_at":86},"413b7c1f-e12b-4571-8076-8b5511360bbd","AlayaWorld开源:用3D缓存+DMD蒸馏破解长时视频世界模型一致性难题","alayaworld-long-video","2026-07-14T10:00:00+00:00"]