prime-rl 0.10.0 发布:训练提速、新 RL 损失 IPO 与统一评测栈
Prime Intellect 发布 prime-rl v0.10.0 与 verifiers v0.4.0,新增 RL 算法 IPO 并设为默认损失。Qwen3-235B-A22B SFT 在 4×8 B300 上比 torchtitan 快 1.15×,MFU 从 40.7% 升至 46.6%。
Prime Intellect 发布 prime-rl v0.10.0 与 verifiers v0.4.0,新增 RL 算法 IPO 并设为默认损失。Qwen3-235B-A22B SFT 在 4×8 B300 上比 torchtitan 快 1.15×,MFU 从 40.7% 升至 46.6%。
Pumpkin Bot is coming (for a limited time)
来自 @ollama 的 Day0 支持 🦙🤝🐰 M3.1 Flash,美国托管,零数据保留,已为你的智能体准备就绪。开始构建吧!
MiniMax M3.1 Flash is on Ollama! 🔒 Hosted in the US and with zero data retention. Fast coding and agentic workflows, with image understanding and a 1M-token context window. Try it on Ollama Cloud: ollama run minimax-m3.1-flash:cloud Claude Code: ollama launch claude --model minimax-m3.1-flash:cloud Codex: ollama launch codex -–model minimax-m3.1-flash:cloud
Amp 现已作为智能体接入 Linear,把 issue 指派给 Amp 会在 Linear 中启动新的 Agent Session 并生成一个 orb,也可通过 @Amp 触发会话。
Grok Bot is now one step closer. Set it as your iPhone's Action Button and press it to talk with Bot.
AWS 周一发布 TorchNeuron 工具,让基于 PyTorch 的 AI 应用只需改一行配置即可在自家 Trainium 芯片上运行,无需重写代码,直接挑战 Nvidia 凭 CUDA 在 PyTorch 生态中的默认地位。
Grok Bot is now one step closer. Set it as your iPhone's Action Button and press it to talk with Bot.
Eugenia Kuyda 于周一正式向所有用户推出 Wabi 2.0,一款以消息应用形式呈现的多用户个人 AI 助手,会连接邮箱后自动创建线程和应用,用户也可请求定制应用。
Hugging Face 宣布 Transformers、Diffusers、Accelerate、PEFT 和 TRL 通过 TorchNeuron 原生支持 AWS Trainium。
Grok Bot is now one step closer. Set it as your iPhone's Action Button and press it to talk with Bot.
Grok Bot is now one step closer. Set it as your iPhone's Action Button and press it to talk with Bot.
Introducing FLUX 3 Max. Fastest in the west. Fast enough to try anything. And everything. Generate a 10-second video in 12 seconds. Tied for #1 in image-to-video. 22× faster than Seedance 2.5 at the same quality. FLUX 3 is a model designed for customization. @fal adapted and optimized it to create FLUX 3 Max.
Claude 宣布 Dashboards 和 Motion 进入 beta。Dashboards 可连接数据平台或 CRM,生成随数据变化实时更新的仪表盘,面向所有付费计划;Motion 把报告或图表变成动画讲解短片,以代码方式生成、无视频模型参与,可修改文字、数字和节奏后导出 MP4,面向 Team 和 Enterprise 计划。
同一件事,精选收录的是《Claude 推出 Dashboards 实时仪表盘与 Motion 动画讲解功能》
Grok Bot 又近了一步。 把它设为 iPhone 的操作按钮,按下即可与 Bot 对话。
Together AI 推出 Together Post-Training 服务,支持用户自定义完整训练循环,通过 RL 环境奖励和 SFT 数据训练专属模型。
Kling AI 发布视频生成模型 Kling 4.0,单次可生成最长 30 秒视频,提示词理解扩展到 8000 token,Omni Reference 最多接受 15 个参考素材,新增双声道立体声和十余种语言的对口型。
Mastra Studio 现在支持自带认证,团队可用 WorkOS、Okta、Auth0、Clerk 或 Google 登录,并将 Studio 与 server 的认证权限分开配置。
Google 发布 LiteRT Tensor API,一个轻量 C++ 模型编写层,可在 CPU、GPU(ML Drift)、NPU 上运行同一计算图,或经 XNNPACK 无需模型文件直接在 CPU 运行。
The compute market has been a wild ride in 2026. Prices have skyrocketed, minimum contract durations have tripled, and near term capacity is few and far between. At the same time, agents are becoming a primary user of compute themselves, and not just GPU compute. It’s clear that the existing rental models and platforms across the industry no longer fit the reality we see today. That’s why this week we’re announcing six new features for the Shadeform platform 👇️
推荐理由:原文给出 Inception 扩散模型的速度、上下文、基准对比和上线折扣价格,可据此评估其相对现有模型的性价比。
Wabi 2.0 is live! personal agents are one person, one chat. but the parts of life that matter run on more than one person Wabi is an agent you use together. you talk to it, it does stuff, builds the apps you need, and anyone can join the thread - your partner, your best friend, the preschool moms groupchat school changes pickup time → Wabi updates the family calendar, everyone sees it. a show you’d both love → Wabi finds a night that works, gets the tickets. the trip everyone keeps talking about → Wabi picks the dates, builds the itinerary, books it go start a thread with your people 👇
DeepSeek V4.1 Flash 在 OpenCode Go 和 Free 中已更新至 1017 你这边无需做任何改动
Started from a Roadster concept trailer, then built this full-stack game with @ManusAI Game Dev! Game tracks: Golden Gate Bridge, Tokyo Tower and Marina Bay Sands Log in to save progress, see play history, and join the leaderboard Play now: https://flyingcars.manus.game
ChatGPT Images 2.5—faster, sharper, smarter, with better tools for creating whatever you can dream of. - Faster image generation to keep your ideas flowing - Improved fidelity for more natural, recognizable images - Consistent details across multiple edits - Comment-based edits to change only what you want
AWS 宣布推出 TorchNeuron,一个面向 Trainium 的开源原生 PyTorch 后端,10 月 20 日以 Apache 2.0 协议在 PyTorch Conference North America 开启公测。
推荐理由:原文给出 TorchNeuron 的架构、执行模式和兼容性测试数据,开发者可据此评估把现有 PyTorch 工作流迁到 Trainium 的成本。
🔥 现在 ChatGPT 和 Claude 都可以直接使用 Plugins 安装 Podwise 处理 Podcast 信息了!
🥳 Podwise is now available as a ChatGPT Plugin. https://chatgpt.com/plugins/plugin_asdk_app_6abe9890e8f08191b122c4f30a19c956
作者实测 Lovart 2.0 新上线的个人设计 Agent Arti,它能读取用户历史项目的审美记忆,在生成中自动调研视觉方向、自主纠错,并支持画布巡检和分享链接后在画布上直接评论。文中用广东文创品牌和奶茶秋季海报项目验证记忆复用,还试运行了一个每日饮品节点的设计委托任务;Lovart 2.0 中文版已全量上线。
SkyProduction 天工工作台 10 月 19 日正式推出官方资产库,开放 6000 多套人设、场景、道具资产,支持标签筛选、预览和跨项目复用。资产库将一次性素材沉淀为可搜索、可再创作的创作资产,官方称可省下 80% 的素材时间。