蚂蚁 inclusionAI 发布 Ling-3.0-tiny 混合推理 MoE 模型 GGUF 版本
蚂蚁 inclusionAI 在 Hugging Face 发布 Ling-3.0-tiny 的 GGUF 版本。该模型为轻量混合推理 MoE 模型,总参数 7.9B,每 token 仅激活 1.3B 参数,主打低推理成本下的推理与智能体能力,适合本地和资源受限部署;提供 Q6_K(6.5 GB)和 BF16(15.8 GB)量化文件。
蚂蚁 inclusionAI 在 Hugging Face 发布 Ling-3.0-tiny 的 GGUF 版本。该模型为轻量混合推理 MoE 模型,总参数 7.9B,每 token 仅激活 1.3B 参数,主打低推理成本下的推理与智能体能力,适合本地和资源受限部署;提供 Q6_K(6.5 GB)和 BF16(15.8 GB)量化文件。
腾讯今日发布 Hy4 Preview,这是一款新的开源权重纯文本输入(无视觉)大语言模型,总参数 770B,激活参数 49B,上下文窗口 1M token,Hugging Face 上体积 1.56TB。相比 7 月的 Hy3(295B 总参数、21B 激活、256k 上下文、598GB),规模大幅提升。Hy4 的推理配置仅支持 "high"(默认)和 "no_think" 两档。
When I try and explain that tokens inferenced increased 25-fold in a year and are multiplying exponentially, cascading through the economy, and are likely to push real GDP growth per year toward double digit territory, many clients and companies simply do not comprehend.
Tencent HY open-sourced Hy4 preview: a flagship MoE with 1M context, and their biggest generational leap measured to date. Apache 2.0.🔥 🤖 https://modelscope.cn/models/Tencent-Hunyuan/Hy4-preview ⚙️ 770B total params, 49B active per token. Ships with a native MTP layer for speculative decoding out of the box. 📊 Blind eval by 163 internal experts across 203 real engineering tasks: it edges out GLM 5.3 (avg 2.99 vs 2.92) and Kimi K3 (2.99 vs 2.94). 💼 Trained on real production workflows: software engineering, office and data analysis, game development, and scientific research. 🌍 For global users: https://modelscope.ai/models/Tencent-Hunyuan/Hy4-preview
"Long-horizon tasks are still a joke. They do not work, and I do not care what anybody says. Do not show me a stupid evaluation. Do not tell me about some dumb script you ran for 48 hours. Long-horizon tasks are not handled well. They simply do not work." - Chamath at Stanford AI Club "2nd, complex problems also do not work. They are neither addressed nor handled well. Why is this important? If AI develops like any other technology, we are going to experience an initial rise—the hype cycle. Then, we will see a natural contraction because, somehow and somewhere, something is going to fail. We are all going to see this, and then we will enter what is called the “trough of disillusionment.” I think the business and MBA folks will confirm whether that is true. Afterward, you typically see the slow and gradual adoption of the real, final solution. This happened with the internet, and it has happened in many other cases. The problem is that we are spending hundreds of billions, potentially trillions, of dollars trying to figure out how to cross this chasm. So, what do we do? If we do not figure this out, people will reach the trough of disillusionment and say that AI was a joke. I think we need to be able to bring AI into highly complicated environments and make it work. What is my solution? At a very basic level, you need a symbolic space that guides the embedded space." ---- From "techniahqrobot" YouTube channel, (full video link in comment)
国光量超人工智能研究团队推出行业首个量子增强型大模型玄幂(Xenomi)家族,将量子技术引入大语言模型的决策与推理过程。该模型面向科研和学术领域,在科研科普、路由和智能体决策等任务中较主流开源模型展现出更稳定的专业理解、更清晰的判断和更短的决策路径。玄幂已应用于国光量超的量子计算智能实验平台和夸米多智能体系统。
"Long-horizon tasks are still a joke. They do not work, and I do not care what anybody says. Do not show me a stupid evaluation. Do not tell me about some dumb script you ran for 48 hours. Long-horizon tasks are not handled well. They simply do not work." - Chamath at Stanford AI Club "2nd, complex problems also do not work. They are neither addressed nor handled well. Why is this important? If AI develops like any other technology, we are going to experience an initial rise—the hype cycle. Then, we will see a natural contraction because, somehow and somewhere, something is going to fail. We are all going to see this, and then we will enter what is called the “trough of disillusionment.” I think the business and MBA folks will confirm whether that is true. Afterward, you typically see the slow and gradual adoption of the real, final solution. This happened with the internet, and it has happened in many other cases. The problem is that we are spending hundreds of billions, potentially trillions, of dollars trying to figure out how to cross this chasm. So, what do we do? If we do not figure this out, people will reach the trough of disillusionment and say that AI was a joke. I think we need to be able to bring AI into highly complicated environments and make it work. What is my solution? At a very basic level, you need a symbolic space that guides the embedded space." ---- From "techniahqrobot" YouTube channel, (full video link in comment)
An internal version of Astra, @OpenAI’s next major model family, solved 10 major open problems in mathematics, quantum complexity, and theoretical computer science. We believe it will be a major step for scientific reasoning. https://openai.com/index/ten-advances-in-mathematics/
Z.ai 与 Qwen 各自独立推出了近乎相同的架构:3:1 线性混合、压缩索引器、门控残差与 Muon 训练。两家中国 AI 实验室在未互通的情况下,于模型设计上实现了趋同。
Infer-forge 是一套围绕 SGLang 推理优化的内部工程系统,通过 MonoRepo、Harness、Task Loop 与 Task Graph 四种结构,将部署点约束链(模型、SLO、拓扑、运行时、加速平台)转化为可复现、可审计的工程流程。
苹果机器学习研究团队提出一种新方法,将 LLM 视为信息处理规则,利用其与贝叶斯更新的信息处理差距,研究模型如何根据证据更新概率信念的内部(不)一致性。实验评估了 LLM 在医学、科学、法律等复杂领域的信念更新表现,揭示其概率推理与贝叶斯理想之间的系统性偏差。
GLM-5.3 现已开放权重。 我们最强大的智能体编码与网络防御模型,现已开放下载、运行和定制。 权重:https://huggingface.co/zai-org/GLM-5.3 技术博客:https://z.ai/blog/glm-5.3
AMD 发布 GPU 加速计算软件堆栈 ROCm 10.0.0,版本号纪念其诞生 10 周年,重点关注 Instinct、Radeon、Ryzen AI 平台上的 AI 推理、开发者工具与性能分析。
腾讯混元发布新一代旗舰模型 Hy4 preview,总参数 770B、激活参数 49B、上下文长度 1M,现已开源并在腾讯云 TokenHub 和 OpenRouter 上线。
推荐理由:官方盲测中它与 GLM-5.3、Kimi K3 的分差不足 0.1,选型时 1M 上下文和开源部署成本可能比榜单排名更实际。
Introducing PRAXIST Beta, your autonomous research team🚀 Define the objective, constraints, and what success looks like. PRAXIST discovers the path. From there, PRAXIST takes on the experimental research loop. Multiple Research Peers explore competing approaches in parallel, share useful findings, and build on accumulated evidence across experiments and generations. In partner-provided environments, PRAXIST reached a 100% safe-landing rate in a rocket simulation and reduced accumulated error in an industrial SLAM system from 9.37 centimeters to 5.01 centimeters. From robotics control to quantitative finance, PRAXIST has delivered measurable advances across fundamentally different problems by combining a shared core research architecture with domain-specific tools, knowledge, constraints, and evaluators. More approaches explored. Faster iteration. Greater R&D capacity without proportionally increasing specialist headcount.
蚂蚁百灵发布首个金融增强开放模型 Ling-3.0-flash-Fin,延续 Ling-3.0-flash 架构,保持 124B 总参数、5.1B 激活参数,通过金融语料持续预训练、领域后训练和工具使用优化。
Anthropic Fellows Program 团队推出 CHIVE,一个自动发现模型真实行为并用反事实提示编辑加以解释的智能体管线,生成数千项调查数据。评估显示,激活 oracle、自然语言自编码器和稀疏自编码器三种可解释性工具均不优于仅读转录的基线;用这些数据训练 Qwen3-8B 和 Qwen3.5-397B-A17B 预测反事实结果,可泛化到未参与训练的设置。论文与代码已开放。
一位化名“roon”的OpenAI研究员警告,超快AI推理带来现有防护无法应对的安全风险。一个未对齐的模型若以当前最佳系统50倍的速度运行,可能迅速渗透系统,人类响应团队将难以招架。roon认为仅靠监控不够,需要自主检测与关停机制;该评论针对OpenAI新AI芯片发布,其推理速度优于现有硬件。
Z.ai 发布 GLM-5.3-Flash,3200 亿总参数(180 亿激活)、百万 token 上下文窗口,MIT 许可开源。该模型在 Intelligence Index 上得 57 分,接近 GLM-5.3 的 60 分,但每任务成本仅 0.09 美元,约为后者的 7.5 分之一。该模型全程运行于国产 AI 芯片,Z.ai 自研服务软件实现了与 Nvidia GPU 相当的性能。
截至 2026 年 6 月,我国日均词元调用量已突破 500 万亿,中国大模型在全球竞争中位居第一梯队。当前旗舰模型几乎以月为单位更新,竞争焦点转向智能体落地与生态建设,推理算力需求随之爆发。腾讯混元 3 正式版上线第一周,Token 调用量比上一代混元 2 增长 68 倍。
推荐理由:模型更新周期缩短至4到6周,且智能体任务的多轮调用说明推理算力需求上升,对算力产业链的判断需从训练侧扩展到推理侧。
Ai2 宣布与 Providence Swedish Cancer Institute 旗下的 Paul G. Allen Research Center 合作,将 AI 科学发现平台 AutoDiscovery 应用于癌症研究数据集。
小鹏集团通用智能中心负责人刘先明在物理 AI 分享会上表示,第二代 VLA 升级的核心是让模型第一次建立起对时间维度的理解。小鹏物理世界基座模型首次将“时间”纳入模型,AI 理解维度从“3D 空间”跃迁至“4D 时空”,并引入 Infini-VLA 长时序架构,可记住前 30 秒的世界。第二代 VLA 同步采用流式自回归推理,端到端响应提速 300%,实现“边看、边想、边行动”的并行处理。
Qwen3.8-Flash 现已上线 @OpenRouter!⚡ 编程助手、智能体工作流、长视频理解,一个 API 调用即可全部实现。
Qwen3.8 Flash from @Alibaba_Qwen is now live on OpenRouter. A multimodal reasoning model for coding assistants, agentic workflows, visual understanding, codebase and document analysis, desktop interaction, charts, and long video. https://openrouter.ai/qwen/qwen3.8-flash
Ox Alpha = GLM-5.3 Flash AA = 57,以1/100的前沿价格, 由纯国产芯片驱动。在OpenRouter上实现了近20%的周token份额(第一)。 感谢大家的支持。
推荐理由:OpenRouter 周 token 份额近两成说明开发者已在用真实调用投票,低价模型正从价格敏感场景切入,团队评估模型成本时会更多参考实际使用量而非榜单单项分。
美团与北京大学提出GeoRA,将低秩适配显式对齐到RLVR的更新几何,先用谱先验与欧氏先验定位稀疏更新区域,再用SVD压缩为低秩稠密适配器并冻结残差锚点。在1.5B至32B的Qwen与Llama模型上,GeoRA在数学、医学、代码三类RLVR任务上稳定优于LoRA、PiSSA、MiLoRA等基线,Qwen3-8B的AIME24达23.75%,可训练参数较全参微调降低99.5%。
黄仁勋在财报电话会议上回应 OpenAI 自研推理芯片 Jalapeño,称英伟达提供的是覆盖 AI 生命周期的完整 AI 工厂平台,可在任何云环境使用,与针对单一模型的定制化 ASIC 芯片有本质区别。他 100% 确信英伟达技术对 OpenAI 仍将卓越。此前测试显示,Jalapeño 在 DeepSeek R1 上每瓦 AI 吞吐量是 GB300 的 1.7 倍。
湖北星辰技术有限公司完成近50亿元A轮融资,刷新国内先进封装领域融资纪录,规划月产10万片。其二期中试线预计9月全线通线,一期二期合计投资超70亿元,月产能规划2万片,已建成国内规模最大高密度集成封装中试和小批量量产平台。通过先进封装,星辰技术将推理带宽比传统结构提升1到2个数量级,正协助国内头部企业打磨三维计算芯片。
智谱正式发布GLM-5.3-Flash,即此前在OpenRouter和OpenCode上爆火的匿名模型OX Alpha。该模型为总参数320B、激活18B的MoE架构,是GLM-5系列首个原生多模态模型,支持图像、视频和文本输入,原生上下文100万。
英伟达发布 2027 财年半年报,上半年营收 1778.37 亿美元,归母净利润 1180.1 亿美元,同比增长 161.1%,GAAP 毛利率 75%。第二财季营收 962.21 亿美元,同比增长 106%,环比增长 18%,归母净利润 596.88 亿美元,同比增长 126%。数据中心业务第二季度收入 890.23 亿美元,同比增长 117%,Vera Rubin 平台已进入全面量产。
推荐理由:数据中心收入同比翻倍叠加 Vera Rubin 平台全面量产,显示 AI 基础设施的资本开支仍在加速,延迟部署的概率下降。
Agnes AI 发布 Agnes 2.5 Pro Beta,在 Artificial Analysis Intelligence Index 得分 49,比 Agnes 2.5 Pro Alpha 的 40 分提高 9 分,接近 Gemini 3.5 Flash(high,52)和 GPT-5.6 Luna(max,52)。
Z.ai 证实此前神秘模型 Ox Alpha 实为 GLM-5.3-Flash,其每日 100 万亿 token 的流量由数万块国产 AI 芯片支撑,而非 NVIDIA GPU 集群。
OxAlpha is a new iteration of GLM, from China’s Z .ai And it will change how you run long running agent fast. --- bloomberg .com/news/articles/2026-08-26/china-s-z-ai-made-ox-alpha-stealth-model-that-rivals-deepseek
投机解码通过让一个小模型预先草拟多个候选 token,再由大模型在单次前向传播中并行验证,可将生成速度提升 2-3 倍,且输出在统计上与单独运行大模型完全一致。70B 参数模型每生成一个 token 需从 GPU 内存读取约 140 GB 权重,而生成阶段计算利用率仅 20-40%,投机解码正是利用这部分闲置算力。文章还探讨了草稿来源的四种方式及投机解码失效的场景。