Latest

Every published item, in order · 6305 items
Nov 12
Chinese contentPhysical Intelligence π0.5 加入触觉编码器后成功率从 48% 提升至 70%

这个组合很实在:蚂蚁旗下的 LingBot-Touch 触觉编码器只有 5.46M 参数,接进 π0.5 就把 26 项任务成功率从 48% 拉到 70%,还开源了。做机器人操作的同学可以直接上手试,比堆算力划算得多。

kimmonismus@kimmonismusSource
ModelsBuildersMulti-sourceSelected00:43
vLLM-Omni adds Day-0 support for Runway's Praxis-1 robot model

Apache 2.0 weights plus single-H100 qualification make Praxis-1 one of the more accessible open robot policies to self-host, with no cloud dependency for inference. Teams building robot stacks in ASEAN research labs or startups can prototype without multi-GPU budgets, and the OpenPI/Positronic serving paths mean existing manipulation pipelines can swap in the model with modest glue code. Anyone comparing open robot foundation models should weigh it against restricted-license alternatives.

Event
Official accountvLLM@vllm_projectSource
ModelsBuildersMulti-source00:42
Chinese contentFishAudio 发布 Live 3:支持全双工语音的实时对话模型

FishAudio 这个 Live 3 能边听边说,不用像对讲机那样一人一句,语音智能体的打断体验会自然很多。vLLM 团队还主动喊话说开源后第一时间支持。做语音 Agent 的开发者可以先盯住它的开源动向。

Event
Official accountvLLM@vllm_projectSource
ModelsUS vs ChinaMulti-source86°00:28
Chinese contentQwen4 发布:Qwen4-Max 开启预览,主打编码与智能体能力

Qwen4-Max 现在就能试了,Token Plan 预览通道和 Qoder、QwenStudio 四个入口都开着。做编码和智能体开发的可以拿真实项目跑一轮,跟 Claude、GPT-5 旗舰比比工具调用和多步骤任务的稳定性。

Event
Official account阿里云 Alibaba Cloud@alibaba_cloudSource
ModelsBuildersSelected00:27
AlphaGenome 2 Launches via API for Non-Commercial Research

API-only access means labs can integrate regulatory predictions into pipelines without GPU infrastructure or model hosting, which shifts the cost of entry from engineering to analysis. Computational biologists, university groups, and biotech research teams working on gene regulation can start experimenting immediately under non-commercial terms. Anyone planning applied or commercial genomics work should check DeepMind's licensing terms first, as the restriction currently blocks revenue-generating use.

Sundar Pichai@sundarpichaiSource
Tomorrow 08:00, the daily lands in your inboxTelegram channel or email. Same screening as this page, no ads.
Chinese contentIFM 开源世界模型 PAN 2,用模拟式推理让智能体先演练再行动

IFM 把世界模型做成了智能体的"预演器":29B 的 LDM 负责画后果,LLM 负责想后果,选最优再动手。和纯文本推理的 agent 框架相比,它多了视频级的可视化验证。做机器人、游戏或决策型智能体的开发者可以直接下载权重和代码上手试。

Ate-a-Pi@svpinoSource
Nov 11
ModelsBuildersMulti-source87°23:40
Chinese contentNVIDIA 发布开源机器人基础模型 GR00T 2,端到端完成化学实验

NVIDIA 把机器人开发全链路都开源了:模型权重、10 万小时数据、仿真引擎、硬件设计全给。演示里 GR00T 2 真手做化学实验还是 1 倍速无剪辑。搞具身智能的团队真该去看看,省多少从头造轮子的功夫。

Event
Official accountJim Fan@jimfanSource
ModelsBuildersMulti-sourceSelected23:35
Runway releases Praxis-1 robot model with open weights

An open-weights, Apache 2.0 robot control model removes the licensing barriers that usually come with robotics foundation models, letting teams self-host and modify freely. Robotics engineers and embodied-AI researchers can fine-tune Praxis-1 on their own hardware instead of licensing closed systems. Compared with closed offerings from rivals such as Physical Intelligence, this gives startups a low-cost starting point for building real-world robot behaviour.

Event
Official accountRunway ML@runwaymlSource
ModelsUS vs ChinaMulti-source23:33
Chinese contentRunway 发布 Praxis-1,速度比 GPT-6 Astra 快 15.6 倍、成本低 51 倍

Runway 把自家机器人模型 Praxis-1 和 GPT-6 Astra 放在同一个工作站上比,结果快了 15.6 倍,每个任务还便宜 51 倍。做机器人或自动化的开发者可以关注这种专用模型和通用 LLM 的选型对比。

Event
Official accountRunway ML@runwaymlSource
ModelsBuildersSelected23:27
AlphaGenome 2 predicts cell-specific DNA reading patterns

The interesting shift is that the model's value comes from context — the same variant can be scored across many cell types, enabling tissue-aware interpretation of disease-linked variants that single-context tools miss. Computational biologists, genetics labs and variant-interpretation teams can start experimenting immediately through the free Colab tutorials without infrastructure setup. Commercial users are excluded from the current licence, so biotech teams should treat it as a research preview and watch how DeepMind prices commercial access later.

Official accountGoogle DeepMind@GoogleDeepMindSource
ModelsUS vs ChinaMulti-source23:24
Chinese contentQwen4-Max 预览版现身:主打 3D 建模、世界知识与 Agentic 编程

Qwen 系列疑似又出新预览版,这次瞄准 3D 建模和 Agentic 编程。我还没看到具体跑分,所以准备这周自己上手测测,尤其想看看它的编程智能体能力比 Qwen3-Max 强多少。想第一时间了解的可以关注后续评测。

Event
向阳乔木@vista8Source
ModelsUS vs ChinaSelected21:30
Baidu Uses RSI to Optimize ERNIE Inference Performance

The acceptance gate — speed gain plus unchanged output — is what turns autonomous code rewriting into something deployable on production inference stacks. Teams running inference servers can borrow the same bounded-scope, verify-before-merge pattern without adopting RSI wholesale. The 6.24x latency result was measured on Baidu's Kunlunxin P800, a domestic Chinese chip, making it a data point for comparing self-optimizing inference against US-lab approaches like NVIDIA-tuned stacks.

小互@imxiaohuSource
ModelsUS vs ChinaMulti-source71°19:12
Chinese content百度发布文心5.5 Flash Preview:280B 参数仅激活 14B,Apache 2.0 开源

百度这次直接把 280B 的新模型按 Apache 2.0 开源,激活 14B 就能跑到 200 tokens/s,还支持 vLLM 和 SGLang 部署,对做私有化部署的团队很友好。对比 DeepSeek 和 Qwen 的开源路线,百度这次在 Agent 任务和知识工作场景下了重注。想验证实际水平的,OpenRouter 上能免费用。

Event
向阳乔木@vista8Source
Chinese contentClaude Fable 5.2 连续生成 76 个不重复 HTML 文件,成本约为 Opus 5.5 的 2 倍

有人拿 Claude Fable 5.2 狂做 76 个网页,游戏、互动实验全都有,一个没坏,前端能力确实压过 Opus 5.5。但注意这波烧掉了两周用量,价格还是 Opus 的 2 倍。做前端 demo 的朋友可以直接去看开源文件找灵感。

AI Will@FinanceYF5Source
ModelsBuildersMulti-source17:44
Chinese contentFable 5.2 发布后创作者案例集中展示:4 支影像无实拍镜头

整理了 4 支全部用 Fable 5.2 生成的短片,7330 颗露珠、五千年书写史、动漫 MV、星战混战都在里面,没一个镜头实拍。想看视频模型实际成片水平的,直接刷这组案例比看官方演示更真实。

Event
AI Will@FinanceYF5Source
ModelsBuildersMulti-source16:45
Chinese contentFable 5.2 五个演示:从骑自行车的鹈鹕到可被 AI 操作的音乐工作站

Fable 5.2 放出了 5 个新演示,鹈鹕骑车、蒸汽火车变机器人这些看脑洞,蜂蜜滴落那段看材质和光影的功力。最特别的是那个音乐工作站,画面里的东西能被 AI 直接操作,不只是出视频。做视频内容或交互产品的可以对照看看自己的场景能不能这么生成。

Event
AI Will@FinanceYF5Source
ModelsBuildersMulti-source16:14
Chinese contentGenspark 与 Baseten 联合后训练开源模型,深度研究能力超过 Opus 5.5

Genspark 和 Baseten 把 Qwen3.5-397B 后训练成深度研究专用模型,内评测分 0.759,直接超过 Opus 5.5 的 0.738,现在已经跑在自家 Deep Research 产品里。做垂直场景落地的团队可以看看这条路:开源基座加定向后训练,单任务能打赢闭源旗舰。

Event
Eric Jing@ericjing_aiSource

Only the latest 2000 items are listed; for older ones see the daily archive