Who remembers this gem https://twitter.com/outsource_/status/2046428978261471484/photo/1
中文: 谁还记得这个瑰宝
Come chat with us 👋🏻 https://discord.gg/dBw4QnDw https://twitter.com/outsource_/status/2046428539499487654/photo/1
中文: 来和我们聊天吧👋🏻
Late night convos about AI & Agents > everything else Discord is definitely the move! Glad y'all voted ✅ Doing a giveaway in Discord this week 👇🏻 https://twitter.com/outsource_/status/2046427571831607680/photo/1
中文: 深夜谈论人工智能与安普;代理与特工;其他一切 Discord 绝对是个转会! 很高兴你们都投了✅ 本周在Discord上进行赠品👇🏻
RT @outsource_: How to install Hermes-Workspace web app: TIP OF THE DAY 💡 1. Click the three dots ⋮ (top right) 2. Go to Cast, save & sha…
中文: RT @outsource_:如何安装 Hermes-Workspace 网页应用: 一天的提示 💡 1。单击三个点 ⋮(右上) 2。前往演员阵容,拯救安普;莎·
RT @outsource_: My AI agent Aurora now lives in a Discord for builders... Free to join. Free guides. Free to talk to her. 📖 6 guides: why…
中文: RT @outsource_:我的人工智能代理Aurora现在为构建者提供了一个 Discord 生活...... 免费加入。免费指南。可以自由地与她交谈。 📖 6 个指南:为什么......
If you're a builder this is where you need to be
中文: 如果你是建筑工人,这就是你需要去的地方
If you're a builder this is where you need to be
中文: 如果你是建筑工人,这就是你需要去的地方
If you're a builder this is where you need to be
中文: 如果你是建筑工人,这就是你需要去的地方
If you're a builder this is where you need to be
中文: 如果你是建筑工人,这就是你需要去的地方
Took building in public to discord!!
中文: 在公共场合建楼引发分歧!!
My AI agent Aurora now lives in a Discord for builders... Free to join. Free guides. Free to talk to her. 📖 6 guides: why agents, how to harness them, etc. 🤖 Aurora on-call 24/7 🏆Weekly builder calls 🥇 1:1 mentorship for those interested Join for free and start building… https://twitter.com/outsource_/status/2046393766399914264/photo/1
中文: 我的人工智能经纪人Aurora现在为构建者提供了一个Discord... 免费加入。免费指南。可以自由地和她说话。 📖 6 指南:为何使用代理、如何驾驭它们等 🤖 奥罗拉全天候待命 🏆 每周建设者电话 🥇 为感兴趣的人提供1:1的指导 免费加入并开始建设......
How to install Hermes-Workspace web app: TIP OF THE DAY 💡 1. Click the three dots ⋮ (top right) 2. Go to Cast, save & share 3. Click Install page as app Should I release the .dmg/.exe Desktop App? https://twitter.com/outsource_/status/2046374181512556678/photo/1
中文: 如何安装 Hermes-Workspace 网页应用: 一天的提示 💡 1。单击三个点 ⋮(右上) 2。前往演员阵容,保存和分享 3。将安装页面作为应用程序点击 我应该发布 .dmg/.exe 桌面应用程序吗?
Glad you approve ✅
中文: 很高兴你同意✅
RT @outsource_: 🏆Level Complete -- https://hermes-workspace.com/ v2 LIVE Tell your HermesAgent -> clone Hermes-Workspace Features: 🤖 Hermes Age…
中文: RT @outsource_:🏆 完整版 - v2 实时版 告诉你的HermesAgent -> 克隆Hermes-Workspace 特点: 🤖 爱马仕年龄......
🏆Level Complete -- https://hermes-workspace.com/ v2 LIVE Tell your HermesAgent -> clone Hermes-Workspace Features: 🤖 Hermes Agent Integration — Direct gateway connection with real-time SSE streaming 🎨 8-Theme System — Official, Classic, Slate, Mono each with light and dark… https://twitter.com/outsource_/status/2046353936483786895/video/1
中文: 🏆 完整版—— v2 实时 告诉你的HermesAgent -> 克隆Hermes-Workspace 特点: 🤖 爱马仕代理集成——通过实时SSE流媒体实现直接网关连接 🎨 8 主题系统——每款有亮暗的官方、经典、单板和单体系统......
USE THIS CODE NOW FOR FREE $10 🤯
中文: 立即免费使用此代码 10美元 🤯
🚨 Moonshot’s Kimi K2.6 is INSANE It went full autonomous agent mode: ->Downloaded Qwen3.5-0.8B locally on a Mac M3 Max ->Coded + optimized a complete inference engine in Zig (a language it barely knew) ->12+ hours, 4,000+ tool calls, 14 iterations Result? Inference speed…
中文: Moonshot 的 Kimi K2.6 是 INSANE 它采用了全自主代理模式: ->在Mac M3 Max本地下载 Qwen3.5-0.8B ->Coded + 优化了 Zig 的完整推理引擎(一种几乎鲜为人所知的语言) ->12 小时以上,4000 多个工具调用,14 个版本 结果?推断速度......
RT @outsource_: What a Monday so far... 🤯Lovable LEAK 🤖Kimi 2.6 👥Kimi Agent 🚨Qwen 3.6 MAX Preview 🚀GPT 5.5 Spud Releasing SOON 😉Herems-Wo…
中文: RT @outsource_:到目前为止,周一...... 🤯 可爱的泄漏 🤖 基米 2.6 👥 基米探员 🚨Qwen 3.6 MAX 预览版 🚀 5.5 喷嘴 发布时间 😉 赫雷姆斯-沃......
RT @outsource_: 🚀 GEMMA 4 JUST WENT SUPER SAIYAN & UNCENSORED 🔥 In case you missed it: @songjunkr dropped SuperGemma4-e4b-abliterated and…
中文: RT @outsource_:🚀 GEMMA 4 刚刚退出了超级赛亚安和潘姆;未审查 🔥 如果你错过了: @songjunkr 删除了 SuperGemma4-e4b 和......
RT @outsource_: NEW 🤯 GLM+ QWEN 18B RUNS ON CONSUMER GPU IT BEATS 35B MoE AT HALF THE VRAM @KyleHessling1 just dropped the healed Qwopus…
中文: RT @outsource_:在消费者显卡上运行全新🤯 GLM+ QWEN 18B 它在半个VRAM下击败35B MoE @KyleHessling1 刚刚将愈合的 Qwopus 放下了......
RT @outsource_: 🤯QWEN 3.6 35B-A3B IS INSANE AT CODING @unslothai recently dropped Qwen3.6-35B-A3B-GGUF The first open Qwen3.6 model and it…
中文: RT @outsource_:🤯QWEN 3.6 35B-A3B 防线 @unslothai 最近放弃了 Qwen3.6-35B-A3B-GGUF 的首款开放式 Qwen3.6 型号,且型号为 ...
RT @outsource_: 🤯GEMMA 4 E4B IS FULLY OBLITERATED @elder_plinius dropped gemma-4-E4B-it-OBLITERATED one of the most aggressive uncenso…
中文: RT @outsource_:🤯GEMMA 4 E4B 已完全取消 @elder_plinius 放弃了 gemma-4-E4B-it-IBLITERATED,这是最具攻击性的 uncenso 之一......
Rapid fire -- complete. Posting benchmarks of all next ✅
中文: 快速火——彻底。 发布所有下一步基准 ✅
🤯GEMMA 4 E4B IS FULLY OBLITERATED @elder_plinius dropped gemma-4-E4B-it-OBLITERATED one of the most aggressive uncensoring jobs on Google’s new 4B model. This version is next level 🤯Refusals dropped from 98.8% → 0% hard refusal ✅Guardrails surgically removed (21/42… https://twitter.com/outsource_/status/2046309750468411567/photo/1
中文: 🤯 GEMMA 4 E4B 完全已取消 @elder_plinius 放弃了 gemma-4-E4B-it-IBLITERATED,这是谷歌新款 4B 模式中最具攻击性的非审查性工作之一。 这个版本是下一个级别 🤯 拒绝率从98.8%下降 → 0% 严厉拒绝 ✅ 通过手术移除的护栏(21/42...
🤯QWEN 3.6 35B-A3B IS INSANE AT CODING @unslothai recently dropped Qwen3.6-35B-A3B-GGUF The first open Qwen3.6 model and it’s solid. This MoE beast is cooking on benchmarks: 🧠SWE-bench Verified: 73.4 (big jump) 🚀 Terminal-Bench 2.0: 51.5 🌟LiveCodeBench v6: 80.4 ✅MMLU-Pro:… https://twitter.com/outsource_/status/2046308640265801773/photo/1
中文: 🤯 编码时,QWEN 3.6 35B-A3B 无 @unslothai 最近放弃了 Qwen3.6-35B-A3B-GGUF,这是首款开放式 Qwen3.6 型号,且表现稳健。 这款MoE兽正以基准进行烹饪: 🧠 SWE-bench 已验证:73.4(大幅跃升) 🚀 终端-台式 2.0:51.5 🌟 LiveCodeBench v6:80.4 ✅MMLU-Pro:......
NEW 🤯 GLM+ QWEN 18B RUNS ON CONSUMER GPU IT BEATS 35B MoE AT HALF THE VRAM @KyleHessling1 just dropped the healed Qwopus-GLM-18B-Merged-GGUF Insane 64-layer frankenmerge of two elite Qwen3.5-9B finetunes (Opus reasoning + GLM-5.1 distill). This thing is cooking on consumer… https://twitter.com/outsource_/status/2046306919346786696/photo/1
中文: 全新 🤯 GLM+ QWEN 18B 运行于消费级显卡 它在半个VRAM下击败35B MoE @KyleHessling1 刚刚将愈合的 Qwopus-GLM-18B-Merged-GGUF 击倒 两种优秀的Qwen3.5-9B精华音(Opus 推理 + GLM-5.1 蒸馏),采用全新的 64 层 frankenmerge。 这款产品正在消费者上烹饪......
🚀 GEMMA 4 JUST WENT SUPER SAIYAN & UNCENSORED 🔥 In case you missed it: @songjunkr dropped SuperGemma4-e4b-abliterated and it’s CLEAN. No refusals. Zero capability loss. Actually beats the base model across the board. Why this 8B beast is different : 🤖Release Quality:… https://twitter.com/outsource_/status/2046305430763159816/photo/1
中文: 🚀 GEMMA 4 刚刚战胜了超级赛扬和安普;未接受 🔥 如果你错过了: @songjunkr 将 SuperGemma4-e4b 消除,且已清洁。没有拒绝。零能力损失。 实际上,它全面超越了基础模型。 为什么这个8B型动物与众不同: 🤖 发布质量:
Bout to post rapid fire Open Source trending models from hugging face... Please forgive me But X needs to see all of these 😉
中文: 推出快速启动开源的热门模特,拥抱面容......请原谅我 但X需要看到所有这些😉
Whats more important? Building or Marketing?
中文: 还有什么更重要的?建设还是营销?
Build in public I promise
中文: 在公共场合建造我保证
Growing on X is all about consistency. Ive had this account for like 10 years. Never took it serious until a month ago. Now 10K Posts later Build in public. https://twitter.com/outsource_/status/2046301667914813451/photo/1
中文: 在X上成长完全取决于一致性。 这个账户已经有十年了。 直到一个月前才认真对待。 现在10K帖子之后 在公共场合进行建设。
How to debug and ship hands-free in 2026 🤯 1. User pastes install error 2. Agent finds root cause (pkg not on PyPI, not PEP 668) 3. Rewrite + push + deploy 4. Smoke test green, tweet draft I typed zero commands. https://twitter.com/outsource_/status/2046300640276853218/photo/1
中文: 2026年如何免提调试和发货🤯 1。用户粘贴安装错误 2。代理发现根本原因(PyPI 上不提供 pkg,未在 PEP 668 上) 3。重写 + 推送 + 部署 4.烟雾测试绿色,推文草稿 我输入了零命令。
How to debug and ship hands-free in 2026 🤯 1.user pastes install error 2. agent finds root cause (pkg not on PyPI, not PEP 668) 3. rewrite + push + deploy 4.Smoke test green, tweet draft I typed zero commands. https://twitter.com/outsource_/status/2046300517064933598/photo/1
中文: 2026年如何免提调试和发货🤯 1. 用户粘贴安装错误 2. 代理发现根本原因(PYPI 上未找到 pkg,未在 PEP 668 上使用) 3. 重写 + 推送 + 部署 4.烟雾测试绿色,推文草稿 我输入了零命令。
How to debug and ship hands-free in 2026 🤯 1.user pastes install error 2. agent finds root cause (pkg not on PyPI, not PEP 668) 3. rewrite + push + deploy 4.Smoke test green, tweet draft I typed zero commands. https://twitter.com/outsource_/status/2046300262500098061/photo/1
中文: 2026年如何免提调试和发货🤯 1. 用户粘贴安装错误 2. 代理发现根本原因(PYPI 上未找到 pkg,未在 PEP 668 上使用) 3. 重写 + 推送 + 部署 4.烟雾测试绿色,推文草稿 我输入了零命令。
RT @outsource_: Hermes-Workspace v2 Mobile is 🔥🔥 What theme are you running? Hermes Nous theme is my favorite right now HermesAgent + Wor…
中文: RT @outsource_:Hermes-Workspace v2 Mobile 是 🔥🔥 你在跑步什么主题? 赫梅斯·诺斯主题是我现在最喜欢的 赫梅斯特工 + 工作...
Open Source Benchmarks on 4090 👇🏻 Pulling - > Benchmarking models In BenchLoop Reply any you want to see 👇🏻
中文: 4090 上的开源基准 👇[EE] 拉力 - > BenchLoop 基准测试模型 回复任何您想查看的👇🏻
I see this image going around a lot I had to fix it for you guys with a skill issue Opus 4.7 > Opus 4.5 > Opus 4.6 https://twitter.com/outsource_/status/2046286511147266315/photo/1
中文: 我经常看到这张图片 我不得不为那些有技能问题的人解决这个问题 Opus 4.7 > Opus 4.5 > Opus 4.6
What a Monday so far... 🤯Lovable LEAK 🤖Kimi 2.6 👥Kimi Agent 🚨Qwen 3.6 MAX Preview 🚀GPT 5.5 Spud Releasing SOON 😉Herems-Workspace V2
中文: 到目前为止,多么愉快的星期一...... 🤯 可爱的泄漏 🤖 基米 2.6 👥 基米探员 🚨Qwen 3.6 MAX 预览版 🚀 5.5 喷子 发布 😉 雷姆斯-工作空间 V2
ANYONE SPARE VRAM ?? NEED MORE GPU 😂 Elon is right, we need more energy. We need unlimited energy production and chips advancing faster. https://twitter.com/outsource_/status/2046284888664371375/photo/1
中文: 有谁的SPARE VRAM ?? 需要更多 GPU 😂 埃隆说得对,我们需要更多精力。 我们需要无限量的能源生产和芯片加速发展。
RT @outsource_: 🚀 FRONTIER 1T MoE on your RTX 4090?! Moonshot’s Kimi K2.6 just dropped… and @UnslothAI already quantized it. Yes. You…
中文: RT @outsource_:🚀 Frontier 1T MoE 在您的RTX 4090上?! Moonshot 的 Kimi K2.6 刚刚下落......以及 @UnslothAI 已经量化了它。 是的。你......
Kimi launched there own agent. It’s happening in front of your face The future is clear. Model + Harness OpenSource will win.
中文: 基米在那里创办了自己的经纪人。 事情正发生在你面前 未来是明确的。 型号 + 内用 开源将获胜。
I can’t tweet. It’s broken. GGS
中文: 我不能发推特。它坏了。GGS
🚀 FRONTIER 1T MoE on your RTX 4090?! Moonshot’s Kimi K2.6 just dropped… and @UnslothAI already quantized it. Yes. You can run a full 1 trillion parameter thinking/agentic beast locally on consumer hardware. Here’s how: ✅ Grab the Unsloth Dynamic UD-TQ1_0 (1.8-bit) →… https://twitter.com/outsource_/status/2046282233905115417/photo/1
中文: 🚀 在您的RTX 4090上使用Frontier 1T MoE?! Moonshot 的 Kimi K2.6 刚刚下落......以及 @UnslothAI 已经量化了它。 是的。你可以在本地消费硬件上运行一个完整的1万亿个思维/代理式参数。 如下: ✅ 获取 Unsloth Dynamic UD-TQ1_0(1.8 位) →
If you’re NEW say Hi 👋🏼 Reply for a follow back 👇🏻👇🏻 https://twitter.com/outsource_/status/2046276355067687126/photo/1
中文: 如果你是新人,请说你好👋🏼 回复回复 👇🏻👇🏻
We're in a DIGITAL renaissance. Get your hand dirty with: Old gaming PCS, Old MacBooks. Unified Memory ✅Nvidia RTX ✅ Hardware cannot match the demand. Software is advancing exponentially. Ram prices are at all time highs. Ram shortages are REAL. https://twitter.com/outsource_/status/2046273911394808009/video/1
中文: 我们正处于数字复兴之中。 用以下方式弄脏你的手: 旧游戏PCS,旧款MacBook。 统一内存 ✅ Nvidia RTX ✅ 硬件无法满足需求。 软件正在呈指数级增长。 公羊价格一直很高。 公羊短缺是真实存在的。
We're in a DIGITAL renaissance. Get your hand dirty with: Old gaming PCS, Old MacBooks. Unified Memory ✅Nvidia RTX ✅ Hardware cannot match the demand. Software is advancing exponentially. Ram prices are at all time highs. Ram shortages are REAL. https://twitter.com/outsource_/status/2046273576018256032/video/1
中文: 我们正处于数字复兴之中。 用以下方式弄脏你的手: 旧游戏PCS,旧款MacBook。 统一内存 ✅ Nvidia RTX ✅ 硬件无法满足需求。 软件正在呈指数级增长。 公羊价格一直很高。 公羊短缺是真实存在的。
RT @outsource_: NEW Operations Feature in https://hermes-workspace.com/ 🏆Operations - Your persistent agent team 🧠Configure sub agents & schedu…
中文: RT @outsource_: 新的操作功能,网址为 🏆 操作——您的持久代理团队 🧠 配置子代理和样品;示
RT @outsource_: NEW Conductor Feature in https://t.co/JflLUfop4O🤯 🤖Conductor - Launch a mission and watch your agent team build it live.…
中文: RT @outsource_:新导约功能,网址为 🤖 导管 - 启动任务,并观看您的代理团队实时构建任务。......
Hermes-Workspace is approaching -> 2,000 ⭐️ Feature List: 📲Dashboard +Mobile 💻Chat + Sessions ⚙️Jobs + Tasks 🖥️Terminal + Files ✅Memory + Profiles 🚀Skills Browser 🔥NEW Operations + Conductor What other features should we add? https://twitter.com/outsource_/status/2046271070919881147/photo/1
中文: 爱马仕-工作空间即将接近 -> 2000 ⭐️ 功能列表: 📲 仪表板 +移动 💻 聊天 + 会话 ⚙️ 工作与任务 🖥️ 终端 + 文件 ✅ 记忆 + 个人资料 🚀 技能浏览器 🔥 新运营与指挥 我们还应该添加哪些其他功能?
Hermes-Workspace is approaching 2K ⭐️ Feature List: 📲Dashboard +Mobile 💻Chat + Sessions ⚙️Jobs + Tasks 🖥️Terminal + Files ✅Memory + Profiles 🚀Skills Browser 🔥NEW Operations + Conductor What other features should we add? https://twitter.com/outsource_/status/2046271010601586996/photo/1
中文: 爱马仕-工作区正在接近2K ⭐️ 功能列表: 📲 仪表板 +移动 💻 聊天 + 会话 ⚙️ 工作与任务 🖥️ 终端 + 文件 ✅ 记忆 + 个人资料 🚀 技能浏览器 🔥 新运营与指挥 我们还应该添加哪些其他功能?
Hermes-Workspace is approach 2K ⭐️ Feature List: 📲Dashboard +Mobile 💻Chat + Sessions ⚙️Jobs + Tasks 🖥️Terminal + Files ✅Memory + Profiles 🚀Skills Browser 🔥NEW Operations + Conductor What other features should we add? https://twitter.com/outsource_/status/2046270970814513562/photo/1
中文: 爱马仕-工作空间是接近2K的⭐️ 功能列表: 📲 仪表板 +移动 💻 聊天 + 会话 ⚙️ 工作与任务 🖥️ 终端 + 文件 ✅ 记忆 + 个人资料 🚀 技能浏览器 🔥 新运营与指挥 我们还应该添加哪些其他功能?
Hermes-Workspace v2 Mobile is 🔥🔥 What theme are you running? Hermes Nous theme is my favorite right now HermesAgent + Workspace = You’re GMI https://twitter.com/outsource_/status/2046263727276675420/photo/1
中文: Hermes-Workspace v2 Mobile 是 🔥🔥 你在跑步什么主题? 赫梅斯·诺斯主题是我现在最喜欢的 HermesAgent + Workspace = 您是 GMI 的
🚨MYTHOS CONFIRMED LIVE HACKING COMPETITORS
中文: 🚨 MYTHOS 确认了实时的 Hacking 竞争对手
NEW Conductor Feature in https://t.co/JflLUfop4O🤯 🤖Conductor - Launch a mission and watch your agent team build it live. 👥Configure - Orchestrator Model & Worker Agents ✅Auto Mode - For autonomous workflows https://twitter.com/outsource_/status/2046256818620350887/photo/1
中文: 新导体功能,网址为 🤖 导体 - 启动任务,并观看您的代理团队实时构建。 👥 配置 - 管弦乐模型和放大器;工人代理 ✅ 自动模式 - 适用于自主工作流程
NEW Operations Feature in https://hermes-workspace.com/ 🏆Operations - Your persistent agent team 🧠Configure sub agents & scheduled jobs 🌟Chat with orchestrator + sub agents ✅View agent outputs all in on place! https://twitter.com/outsource_/status/2046256807467769884/photo/1
中文: 新的运营功能: 🏆 操作——您的持久代理团队 🧠 配置子代理和职位安排 🌟 使用编排器和子代理进行聊天 ✅ 查看代理输出全部到位!
NEW KIMI K2.6 Lets gooo 🚀 What a time to be in OpenSource
中文: 新基米 K2.6 放手 🚀 在开源中应该是什么时间
🚨 QWEN3.6-MAX-PREVIEW JUST DROPPED And it competes with SOTA Models. China is closing the gap FAST. Try it 👇🏻
中文: 🚨 QWENE3.6-MAX-PREVIE 刚刚已关闭 它与SOTA模型竞争。 中国正在快速缩小这一差距。 试试👇🏻
Most people are WRONG about Opus 4.7 Heres why: Opus 4.7 is very autonomous & great at coding. Given the correct tools and context Feed it simple prompts and it runs for 30+ mins No user interaction. It gets as much context it needs and runs. Most people are completely… https://twitter.com/outsource_/status/2046232417649004557/photo/1
中文: 大多数人对《行动4.7》持错误态度 原因如下: Opus 4.7 非常自主,非常擅长编程。 给出正确的工具和上下文 简单提示,持续30多分钟 无用户交互。 它获得所需的大量背景和运行。 大多数人完全是......
I have a bunch of OpenSource Models to test today👇🏻 https://huggingface.co/KyleHessling1/Qwopus-GLM-18B-Healed https://huggingface.co/hesamation/Qwen3.6-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled https://huggingface.co/mlx-community/Qwen3.6-35B-A3B-4bit-DWQ + More! Will benchmark + post Reply any others you're curious about👇🏻 https://twitter.com/outsource_/status/2046217562304954651/photo/1
中文: 今天我有大量开源模型需要测试👇🏻 + 更多! 将进行基准测试并发布 回复任何你感兴趣的人👇
Happy Monday! Let’s get cranking 🚀 Releasing discord link today. Chat with my agent for help Get free guides on how to use Agents to ship Weekly calls etc.
中文: 周一快乐! 让我们开始调整 🚀 今天就释放不和联系。 与我的经纪人聊天寻求帮助 获取如何使用 Agents 发货的免费指南 每周通话等
RT @outsource_: New website goes 🔥🔥 10,000+ Clones 1600+⭐️ https://hermes-workspace.com/ ⭐️ https://twitter.com/outsource_/status/2046080185548636565/photo/1
中文: RT @outsource_:新网站发布 🔥🔥 10000 多克隆 1600+⭐️ ⭐️
RT @outsource_: Introducing https://hermes-workspace.com/ V2 🤯 Whats New 👇🏻 🤯No fork required ⭐️New Hermes dark + light themes 🤖Agent View Off…
中文: RT @outsource_:介绍 V2 🤯 什么新的👇🏻 🤯 无需分叉 ⭐️ 新爱马仕暗黑+光主题 🤖 探戈 查看...
New website goes 🔥🔥 10,000+ Clones 1600+⭐️ https://hermes-workspace.com/ ⭐️ https://twitter.com/outsource_/status/2046080185548636565/photo/1
中文: 新网站 🔥🔥 10000 多克隆 1600+⭐️ ⭐️
Introducing https://hermes-workspace.com/ V2 🤯 Whats New 👇🏻 🤯No fork required ⭐️New Hermes dark + light themes 🤖Agent View Office 🚀Conductor for agent missions 👥Operations for sub agent orchestration One liner: curl -fsSL https://hermes-workspace.com/install.sh | bash https://twitter.com/outsource_/status/2046079580105064787/video/1
RT @outsource_: New Themes for https://hermes-workspace.com/ And new release 😉 Getting everything wrapped up https://twitter.com/outsource_/status/2045997337730826711/photo/1
中文: RT @outsource_: 的新主题 以及新版本 😉 把所有东西都收起来
How to Ship in 2026: Prompt your agent That’s all. That’s all you need to do It’s so easy to ship. These models are so capable Given the right tools and harness They can do anything you ask them to. Get creative. Have them rewrite your life. https://twitter.com/outsource_/status/2046050887462694930/photo/1
中文: 2026年如何发货: 提示你的代理人 就是全部。 这就是你需要做的 发货非常方便。 这些模型非常有能力 赋予正确的工具和工具 他们可以做任何你要求他们做的事情。 发挥创意。让他们重写你的生活吧。
How it feels when you install HermesAgent ⭐️ curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash https://twitter.com/outsource_/status/2045998581849415988/video/1
中文: 安装 HermesAgent 时的感觉如何 ⭐️ 卷发 -fsSL | bash
New Themes for https://hermes-workspace.com/ And new release 😉 Getting everything wrapped up https://twitter.com/outsource_/status/2045997337730826711/photo/1
中文: 的新主题 以及新版本 😉 把所有东西都收起来
Nobody: Claude: https://twitter.com/outsource_/status/2045885735593451768/photo/1
中文: 没人: 克劳德:
It only makes sense. As the claw crashes HermesAgent takes over ⭐️ https://twitter.com/outsource_/status/2045859750085050435/photo/1
中文: 这才有意义。当爪子崩溃时 HermesAgent 接管 ⭐️
RT @Teknium: Hermes Agent just hit 100,000 stars on GitHub!!! Thank you everyone!! https://twitter.com/Teknium/status/2045739848913600640/photo/1
中文: RT @Teknium:爱马仕代理在 GitHub 上刚刚登上了十万颗星! 谢谢大家!
RT @outsource_: I use OpenClaw AND HermesAgent side by side 👇🏻 Ask me ANYTHING 👋🏻 Most OpenClaw setups a fragile and break. Mines develop…
中文: RT @outsource_:我并排使用 OpenClaw 和 HermeSagent 👇🏻 问我什么 👋🏻 大多数OpenClaw会设置一个脆弱且断裂的装置。 矿山正在发展......
RT @outsource_: My agent is cooking for Hermes Agent Creative Hackathon🤯 I asked my agent to get creative... I expected AI slop. From one…
中文: RT @outsource_:我的经纪人正在为爱马仕经纪人创意黑客马拉松做烹饪🤯 我让我的经纪人发挥创意...... 我预料到人工智能会倾斜。 从一个......
Buying a RTX 6000
中文: 购买RTX 6000
Buying a GPU
中文: 购买GPU
GROK 4.20 → GROK 4.3 BETA They are making grok more agentic.. I asked it to explain its new capabilities: ⚡ Lightning speed 👥 4 agents 💬 2M context ✅ Lowest hallucinations 📹 Native Video Analysis 📄 Instant PDF / PPT / Spreadsheet Generation 🚀 2× bigger brain (1T… https://twitter.com/outsource_/status/2045724115470840272/photo/1
中文: GROK 4.20 → GROK 4.3 BETA 他们正在让格罗克变得更加特化。 我请它解释一下它的新功能: ⚡ 闪电速度 👥 4名特工 💬 2M 上下文 ✅ 最低的幻觉 📹 原生视频分析 📄 即时 PDF / PPT / 电子表格生成 🚀 2× 更大的大脑(1T...
My agent is cooking for Hermes Agent Creative Hackathon🤯 I asked my agent to get creative... I expected AI slop. From one prompt it : Found The Met’s open archive and composed real museum sculptures through a Karborn-style render pipeline. From this skill it's building an… https://twitter.com/outsource_/status/2045712270328803724/photo/1
中文: 我的经纪人正在为爱马仕经纪人创意黑客马拉松做烹饪🤯 我让我的经纪人发挥创意...... 我预料到人工智能会慢下来。 从一个提示来: 找到了大都会博物馆的开放式档案馆,并通过一条卡伯恩风格的渲染图管道创作了真正的博物馆雕塑作品。 从这项技能来看,它正在构建一个......
RT @LeanLabsMD: @outsource_ Migrating to Hermes
中文: RT @LeanLabsMD:@outsource_ 迁移到爱马仕
What’s everyone working on tonight?
中文: 今晚每个人都在做什么?
Opus 4.7 is a mad scientist. Diabolical. 24hrs straight with it and I'm convinced: Benchmarks don't tell you what a model is actually like It's your harness + setup + capabilities. You can have it do insane things. You just have to start tinkering.. https://twitter.com/outsource_/status/2045701464736313561/photo/1
中文: Opus 4.7 是一位疯狂的科学家。恶魔。 连续24小时,我确信: 基准不会告诉你模型究竟是什么样子 这是您的线束+设置+功能。 你可以让它做疯狂的事情。 你只需要开始修补。
Guys HERMES shipped X API Skill Tell your agent to fire this up 🔥🔥🔥
中文: 盖斯·赫姆斯发布了X API技能 告诉你的代理人要解雇这件事🔥🔥🔥
Guys we've done it! @NousResearch HermesAgent is #1 HermesAgent is #1 TRENDING this week LETS GOOOO 🚀 Follow the movement and clone HermesAgent. And Pair it with https://hermes-workspace.com/ Shipping new update so my fork is not needed. https://twitter.com/outsource_/status/2045594708869550109/photo/1
中文: 我们干的家伙们!@NousResearch HermesAgent 是第1名 HermesAgent 本周将排名第一 利特斯·古鲁 🚀 跟随这一运动,克隆 HermesAgent。 并将其与 配对 发货更新,以免我的叉子被需要。
Someone say Discord??
中文: 有人说是Discord?
I use OpenClaw AND HermesAgent side by side 👇🏻 Ask me ANYTHING 👋🏻 Most OpenClaw setups a fragile and break. Mines developed over 10,000 chats & 4 months. I still utilize both in my workflows for different things. Hermes Is 100% more useful out the box. OpenClaw is solid if you… https://twitter.com/outsource_/status/2045555935620350383/photo/1
中文: 我并排使用 OpenClaw 和 HermeSagent 👇🏻 问我什么 👋🏻 大多数OpenClaw会设置一个脆弱且断裂的装置。 矿山开发了超过一万个聊天和4个月。 我在工作流程中仍然会使用这两种内容来做不同的事情。 爱马仕完全更有用。 如果你...... 很稳固
BenchLoop going crazy 🔥 (still in early development and testing) The standard of benchmarking local models: -Pull new trending models -Chat -Benchmark -Post results Will be able to see what tk/s / quality others get on your hardware! https://twitter.com/outsource_/status/2045528818983604527/photo/1
中文: 长凳疯狂🔥 (仍处于早期开发和测试中) 本地模型基准标准: 推出新趋势模型 聊天 基准 发布结果 将能够查看您的硬件上的 tk/s/ 质量!
BenchLoop going crazy 🔥 (still in early development and testing) The standard of benchmarking local models: -Pull new trending models -Chat -Benchmark -Post results Will be able to see what tk/s / quality others get on your hardware! https://twitter.com/outsource_/status/2045528453227630711/photo/1
中文: 长凳疯狂🔥 (仍处于早期开发和测试中) 本地模型基准标准: 推出新趋势模型 聊天 基准 发布结果 将能够查看您的硬件上的 tk/s/ 质量!
I let Opus 4.7 RUN all night and it absolutely COOKS It shipped NEW updates for upcoming BenchLoop: 🚀 7 benchmark suites rewritten, 93 tasks 🔥 2 new suites from scratch (artifact-verified bug fixing + JSON schema conformance) ✅ scoring rigor layer 🧟‍♂️design system swapped to… https://twitter.com/outsource_/status/2045491546988958089/photo/1
中文: 我让Opus 4.7 运行了整夜,而且完全可以 为即将推出的 BenchLoop 发布了最新动态: 🚀 7 个基准套件已重写,93 项任务 🔥 2 个全新的套件(经过工件验证的错误修复 + JSON 模式一致性) ✅ 评分严谨 🧟 ♂ 系统已切换至......
I let Opus 4.7 RUN all night and it absolutely COOKS It shipped NEW updates for upcoming BenchLoop: 🚀 7 benchmark suites rewritten, 93 tasks 🔥 2 new suites from scratch (artifact-verified bug fixing + JSON schema conformance) ✅ scoring rigor layer 🧟‍♂️design system swapped to… https://twitter.com/outsource_/status/2045491339052171298/photo/1
中文: 我让Opus 4.7 运行了整夜,而且完全可以 为即将推出的 BenchLoop 发布了最新动态: 🚀 7 个基准套件已重写,93 项任务 🔥 2 个全新的套件(经过工件验证的错误修复 + JSON 模式一致性) ✅ 评分严谨 🧟 ♂ 系统已切换至......
What are we shipping today??
中文: 我们今天发货的是什么?
Lets see what Opus 4.7 can do 👀 https://twitter.com/outsource_/status/2045358948375498888/photo/1
中文: 让我们看看 Opus 4.7 能做些什么 👀
RT @outsource_: WE GOT MINIMAX 2.7 LOCALLY w/ FrankenGPU👇🏻 MiniMax M2.7 (229B params) Running locally on consumer GPUs RTX 4090 + RTX 307…
中文: RT @outsource_:我们在 FrankenGPU👇 上本地 2.7 获取了MINIMAX 迷你Max M2.7(229B参数) 在消费类 GPU 上本地运行 RTX 4090 + RTX 307...
RT @outsource_: GUYS FRANKENGPU IS ALIVE 🧟‍♂️ Built a distributed GPU cluster dashboard to manage my consumer hardware setup: 5 Nodes 👇🏻…
中文: RT @outsource_:GUYS FRANKENGPU 已生 🧟 ♂ 构建一个分布式 GPU 集群仪表板来管理我的消费硬件设置: 5个节点 👇[EE]...
Check this out
中文: 查看此
Did we cook chat
中文: 我们做聊天了吗
I got MiniMax 2.7 running on my local hardware cluster: 🧟‍♂️229 BILLION PARAMETERS ON CONSUMER HARDWARE 🌟RTX 4090 + 3070. All Over LAN Via llamacpp RPC From one prompt this morning to: ✅ Woke up the cluster via SSH ✅ Debugged Windows firewall block rules ✅ Hit 90 tok/s on… https://twitter.com/outsource_/status/2045288695972933685/photo/1
中文: 我的本地硬件集群上运行了MiniMax 2.7: 🧟 ♂️ 229亿名消费者硬件参数 🌟 RTX 4090 + 3070。通过 lamacpp RPC 全 局域网 今天上午从一个提示到: ✅ 通过 SSH 唤醒该集群 ✅ 调试的 Windows 防火墙块规则 ✅ 点击90 tok/s 上...
WE GOT MINIMAX 2.7 LOCALLY w/ FrankenGPU👇🏻 MiniMax M2.7 (229B params) Running locally on consumer GPUs RTX 4090 + RTX 3070 via llama.cpp RPC 60.7GB model, IQ1 quantization Now will add other nodes to the mix and increase tk/s to be usable. https://twitter.com/outsource_/status/2045285733494075758/photo/1
中文: 我们在当地与 FrankenGPU👇 进行了 MINIMAX 2.7 迷你Max M2.7(229B参数) 在消费类 GPU 上本地运行 RTX 4090 + RTX 3070 via lama.cpp RPC 60.7GB 型号,IQ1 量子化 现在将添加其他节点以加入混合模式,并增加 tk/s 以实现可用。
GUYS FRANKENGPU IS ALIVE 🧟‍♂️ Built a distributed GPU cluster dashboard to manage my consumer hardware setup: 5 Nodes 👇🏻 Mac M1 Pro · RTX 4090 · RTX 3070 · RTX 4070 Ti · 2070 84GB combined VRAM Real-time topology, built-in chat + benchmarking Running Qwen3.6-35B at 90 tok/s… https://twitter.com/outsource_/status/2045249235050725820/photo/1
中文: 伙计们,弗兰肯瓜还活着 🧟 构建一个分布式 GPU 集群仪表板来管理我的消费级硬件设置: 5个节点 👇 Mac M1 Pro · RTX 4090 · RTX 3070 · RTX 4070 Ti · 2070 84GB 组合 VRAM 实时拓扑结构、内置聊天功能+基准测试 运行 Qwen3.6-35B 的 90 点/天...
No way
中文: 绝不
Hermes launched the second Hackathon Time to start shipping guys 👇🏻
中文: 爱马仕推出了第二场黑客马拉松 是时候开始发货了👇🏻
Do these labs read X and laugh?? Cause Anthropic gets dunked on every release The only thing they ship well Is the model OpenAI gets good PR Codex is Solid GPT 5.4 is good Still improvements on each end but how much of it do they see 😂
中文: 这些实验室会读X并开怀大笑吗? 原因 Anthropic 在每次发布时都会被扣篮 他们唯一能寄得体的就是这个模型 OpenAI 获得良好的公关 代码是可靠的 GPT 5.4 很好 每一端仍有改进,但其中有多少是😂
Don’t use closed source tools. Use open source ones. Route through any model. Add any capabilities and skills you need. @ClawSuite👇🏻 https://t.co/JflLUfop4O👇🏻 https://twitter.com/outsource_/status/2045180226997481672/photo/1
中文: 不要使用闭源工具。 使用开源的。 通过任何模型。 添加您所需的任何技能和技能。 @ClawSuite👇🏻
yooo @boltdotnew @Lovable ggs in the chat??
中文: @boltdotnew @Lovable ggs 在聊天中?
@boltdotnew @Lovable ggs in the chat?
中文: @boltdotnew @Lovable ggs 在聊天中?
Guys its here SuperGemma 4 E4b has released! Check it out 👇🏻
中文: SuperGemma 4 E4b 已发布! 查看它 👇🏻
My agent paired Mac (32gb)-> PC1(4090) + PC2 (3070) Franken GPU setup combining my local Hardware Results: Qwen3.6-35B-A3B (MoE, 3B active) 88.4 tok/s generation 🔥 83.6 tok/s prompt processing 1880 tokens in 21.6s Reasoning/thinking mode working Next steps : CONTINUE… https://twitter.com/outsource_/status/2045149586373701658/photo/1
中文: 我的代理配对了 Mac(32gb)->PC1(4090)+ PC2(3070) Franken GPU 设置,结合我的本地硬件 结果: Qwen3.6-35B-A3B(MoE,3B 活跃) 88.4 代 🔥 83.6 次快速处理 1880年代币(21.6s) 推理模式/思维模式 下一步: 继续......
UPDATE: Swapped from Exo -> LlamaCPP RPC [Results at the end of video] Exo worked great as a visual layer but I noticed it couldn't combine the GPU's for models across Mac/Nvidia My process: Feed my use case into my agent Agent SSH into PC's to run setup and test Agent updates… https://twitter.com/outsource_/status/2045147299429425397/video/1
中文: 更新:从Exo -> LlamacCPP RPC 更换 [视频结尾的结果] Exo 作为视觉层效果很好,但我注意到它无法将 GPU 与 Mac/Nvidia 的型号组合在一起 我的流程:将我的用例输入我的代理 将 SSH 输入到 PC 中以运行设置和测试 代理更新......
These models are so autonomous its insane Pairing my hardware together across LAN Troubleshooting different setups w Exo/ LlamaCPP RCP SSH into all computers Test/ Benchmark / Adjust. This has been running for 1 hr on autopilot I haven't prompted it to do anything but : How… https://twitter.com/outsource_/status/2045143783348621690/photo/1
中文: 这些模型非常自主,其疯狂 将我的硬件在局域网中配对 使用 Exo/LlamaCPP RCP 解决不同设置的问题 进入所有计算机 测试/基准/调整。 这在自动驾驶仪上已经运行了1小时 我还没提示它做任何事,而是: 如何......
NO LOSERS ALLOWED
中文: 不允许任何失败者
No one: NVIDIA CEO:
中文: 没有人: 英伟达首席执行官:
You are the next AI Researcher You’re researching /testing / using it right now You have more power than you think Engineer the things you wish existed Do what others won’t Don’t give up 🚀
中文: 你是下一个人工智能研究员 你正在研究/测试/使用它 你拥有的权力比想象的要多 设计你希望存在的事物 做别人不会做的事 不要放弃🚀
You are the next AI Researcher You’re researching /testing / using it right now You have more power than you think Engineer the things you wish existed Do what other won’t Don’t give up 🚀
中文: 你是下一个人工智能研究员 你正在研究/测试/使用它 你拥有的权力比想象的要多 设计你希望存在的事物 做别人不会做的事 不要放弃🚀
You are the next AI Researcher You’re researching testing and using it right now You have more power than you think Engineer the things you wish existed Do what other won’t Don’t give up 🚀
中文: 你是下一个人工智能研究员 你现在正在研究测试并使用它 你拥有的权力比想象的要多 设计你希望存在的事物 做别人不会做的事 不要放弃🚀
How I Hit 4.5M+ Impressions in 2 weeks on X -Dial in your niche -Reply and engage with everyone you see -Post in trend with your hot takes -Reply to everyone in your tweets -Push for conversation / bookmark worthy -Hot Times 12-6 PM your time zone Besides that I aim for: -250… https://twitter.com/outsource_/status/2045117834397655551/photo/1
中文: 如何在 X 上 两周内达到 4.5M+ 印象 在你的细分市场中拨号 回复并与你所看到的每个人互动 发布热门话题 - 回复您推文中的每个人 值得推荐聊天/书签 时区12至下午6点 此外,我的目标是: -250...
How I Hit 5M+ Impressions in 2 weeks on X -Dial in your niche -Reply and engage with everyone you see -Post in trend with your hot takes -Reply to everyone in your tweets -Push for conversation / bookmark worthy -Hot Times 12-6 PM your time zone Besides that I aim for: -250… https://twitter.com/outsource_/status/2045117658622771663/photo/1
中文: 如何在 X 上 2 周内 达到 5M+ 印象 在你的细分市场中拨号 回复并与你所看到的每个人互动 发布热门话题 - 回复您推文中的每个人 值得推荐聊天/书签 时区12至下午6点 此外,我的目标是: -250...
Good read on new Claude desktop My friend was using it on my PC The model kept thinking it’s IN Claude CLI Would refuse or not make sense Don think it should have been shipped What are yall experiences?
中文: 在新的克劳德桌面上阅读得不错 我朋友在我的电脑上使用它 该模型一直以为是克劳德·克里 拒绝或不有意义 别以为应该发货了 你们有什么经验?
Today In AI NEWS🗞️ 🤖OPUS 4.7 [ Performs like 4.6 at release ] 💻Perplexity Personal Computer [ Mac computer use ] ✅Claude Resets usage [ Opus 4.7 burns usage ] ⚙️Qwen 3.6 OPEN SOURCE [ #1 Model in category ] 🏆GPT Rosalind [ Chemistry/Biology ] 🖥️Codex Computer Use My…
中文: 今日在人工智能新闻中🗞️ 🤖OPUS 4.7 [发布时表现为4.6] 💻 个人电脑的复杂性 ✅ 克洛德 重置使用量 [Opus 4.7 烧伤使用] ⚙️Qwen 3.6 开放源代码 [类别中排名第一的型号] 🏆 加普·罗萨林德 [化学/生物学] 🖥️ 计算机使用 我的......
Today In AI NEWS🗞️ 🤖OPUS 4.7 [ Performs like 4.6 at release ] 💻Perplexity Personal Computer - Mac computer use ] ✅Claude Resets usage [ Opus 4.7 burns usage ] ⚙️Qwen 3.6 OPEN SOURCE [ #1 Model in category ] 🏆GPT Rosalind [ Chemistry/Biology ] 🖥️Codex Computer Use My…
中文: 今日在人工智能新闻中🗞️ 🤖OPUS 4.7 [发布时表现为4.6] 💻 个人电脑复杂性 - 使用 Mac 电脑 ✅ 克洛德 重置使用量 [Opus 4.7 烧伤使用] ⚙️Qwen 3.6 开放源代码 [类别中排名第一的型号] 🏆 加普·罗萨林德 [化学/生物学] 🖥️ 计算机使用 我的......
Today In AI NEWS🗞️ -OPUS 4.7 (performs like 4.6 at release) -Perplexity Personal Computer (Mac computer use) -Claude Resets usage (Opus 4.7 burns usage) -Qwen 3.6 OPEN SOURCE (#1 Open Model in category) -GPT Rosalind (Chemistry/Biology) -Codex Computer Use My Updates 👇🏻…
中文: 今日在人工智能新闻中🗞️ -OPUS 4.7(发布时表现为4.6) 个人电脑(使用Mac电脑) - 克劳德重置使用(使用 Opus 4.7 烧伤) -Qwen 3.6 开放源代码(类别中排名第一的开放模型) -盖普·罗莎琳德(化学/生物学) - 使用计算机代码 我的更新 👇🏻...
Confirmed: Claude FINALLY Reset Limits. Opus 4.7 BURNS so much tokens they had no choice My take its Opus 4.6 thinking = 100. No big deal, just Opus burning more tokens to think. No thanks! Will test soon tho after all these usage/ concerns pass.
中文: 已确认:克劳德最终重置限制。 Opus 4.7 大量代币,他们别无选择 我认为其“Opus 4.6 思维 = 100”。 没什么大不了的,只是Opus会消耗更多的代币来思考。 不,谢谢! 在所有这些使用/顾虑都通过后,将很快进行测试。
Don’t fall for it guys they want you to burn all your usage in your sleep 😂😂😂😂
中文: 别为此而想,他们希望你在睡梦中把所有用法都烧掉😂😂😂
RT @outsource_: QWEN 3.6 35B-A3B GGUF + @UnslothAI 🦥 35B total params, only ~3B active. What this means: 💻Agentic coding & tool-calling o…
中文: RT @outsource_:QWEN 3.6 35B-A3B GGUF + @UnslothAI 共35B个参数,仅 ~3B 活跃。 这意味着: 💻 正文编码;工具调用 o...
Finally a W guys let’s goo
中文: 终于,一个W人,我们好
QWEN 3.6 35B-A3B GGUF + @UnslothAI 🦥 35B total params, only ~3B active. What this means: 💻Agentic coding & tool-calling on steroids 🧠Native 262K context extendable to 1M+ 👀Built-in vision + multimodal reasoning ⭐️Dynamic 2.0 quants max speed +quality VRAM Requirements:… https://twitter.com/outsource_/status/2044824974771269757/photo/1
中文: QWEN 3.6 35B-A3B GGUF + @UnslothAI 共35B个参数,仅 ~3B 活跃。 这意味着: 💻 对类固醇进行工具编程 🧠 原语境可扩展至 1M+ 👀 内置视觉与多模态推理 ⭐️ 动态2.0量子 最大速度 + 质量 VRAM 要求:
GUYS Franken-GPU is LIVE 🤯 Running 4 Seperate PCS + 1 Mac Hardware Specs: PC 1 - 4090 24 GB VRAM ( 32 GB System Ram) PC 2 - 3070 8 GB VRAM (16 GB System Ram) PC 3 - 4070ti 12 GB VRAM (32 GB System Ram) PC 4 - 2070 8 GB VRAM (16 GB System Ram) Mac - M1 PRO 32 GB Unified… https://twitter.com/outsource_/status/2044802545743233437/photo/1
ITS OFFICIAL CHAT NEW SOTA MODEL DROPPED
中文: 官方聊天新SOTA模型已被拉上
Opus 4.7 Benchmarks LEAKED 👇🏻 Below is a direct side by side comparison of: Opus 4.6 -> Opus 4.7 -> Improvement % Heres how it stacks up: ⚙️ SWE-bench Verified (+6–11%) — Fewer failed bug fixes on real code. 🧠 OSWorld (+5–10%) — Fewer stalls on dynamic UI during computer use.… https://twitter.com/outsource_/status/2044788709791592466/photo/1
中文: Opus 4.7 基准测试 👇🏻 以下是以下是以下直接并列对比: 第4.6-&gt条;Opus 4.7 ->改进% 它如何叠加: ⚙️ SWE-bench 已验证(+6–11%)— 实际代码中失败的错误修复更少。 🧠 OSWorld(+5–10%)——在计算机使用过程中,动态用户界面的失速减少......
Its official guys Opus 4.7 is here Hopefully its not Opus 4.6 at launch Announcement soon Benchmarks soon What a day to be in AI. https://twitter.com/outsource_/status/2044784395530182689/photo/1
中文: 其官方官员Opus 4.7 来了 希望发布时不会推出Opus 4.6 即将发布 基准很快 人工智能中的一天该怎样。
Qwen 3.6 Open source release will be overshadowed by Opus 4.7 🥲
中文: Qwen 3.6 开源版本将被 Opus 4.7 🥲 所掩盖
There's sooo many new OpenSource models everyday Whats good? Whats bad? Whats runs on my hardware? How do we test these at scale? These questions need to be answered... So Ive been developing Bench-Loop All-In-One system to Pull->Benchmark->Share Very early mockup below👇🏻 https://twitter.com/outsource_/status/2044780141755167017/photo/1
中文: 每天都有太多新的开源模型 有什么好处?有什么不好的?我的硬件上有什么问题? 我们如何大规模地测试这些? 这些问题需要得到解答...... 所以我一直在开发Bench-Loop 将一体化系统扩展至 Pull->Benchmark->Share 下图:非常早期的模拟[EE],
There's sooo many new OpenSource models everyday Whats good? Whats bad? Whats runs on my hardware? How do we test these at scale? These questions need to be answered... So Ive been develop Bench-Loop All-In-One system to Pull->Benchmark->Share Very early mockup below👇🏻 https://twitter.com/outsource_/status/2044779618649997390/photo/1
中文: 每天都有太多新的开源模型 有什么好处?有什么不好的?我的硬件上有什么问题? 我们如何大规模地测试这些? 这些问题需要得到解答...... 所以我已经发展了Bench-Loop 将一体化系统扩展至 Pull->Benchmark->Share 下图:非常早期的模拟[EE],
There's sooo many new OpenSource models everyday Whats good? Whats bad? Whats runs on my hardware? How do we test these at scale? These questions need to be answered... So Ive been develop Bench-Loop All-In-One system to Pull->Benchmark->Share Very early mockup below👇🏻 https://twitter.com/outsource_/status/2044779481538187598/photo/1
中文: 每天都有太多新的开源模型 有什么好处?有什么不好的?我的硬件上有什么问题? 我们如何大规模地测试这些? 这些问题需要得到解答...... 所以我已经发展了Bench-Loop 将一体化系统扩展至 Pull->Benchmark->Share 下图:非常早期的模拟👇🏻
Omg gonna cook with this adding this to the Bench-Loop
中文: Omg 会用这个添加到 Bench-Loop 的中添加此菜来烹饪
UPDATE: Got 3 Computers wired up now More VRAM will be available once I get all 5 paired Should work good honestly If this works ill be able to get almost 100gb VRAM. Lets gooo 🚀 https://twitter.com/outsource_/status/2044640881764884588/photo/1
中文: 更新:现已接入3台电脑 更多VRAM将在我配对后全部获得 应该诚实地工作 如果这无法得到近100克的VRAM。 请访问 🚀
Guys…. We have Opus 4.7 https://twitter.com/outsource_/status/2044624993959489721/photo/1
中文: 伙计们......我们有 Opus 4.7
Need GPU so Ive decided to try setup EXO. Setting up EXO on my local network wish me luck 4090 + 4070 + 3070 + 2070 + Mac M1 Pro 32GB Let's see how this goes, if it will fail etc. Let me know what models to test 👇🏻 https://twitter.com/outsource_/status/2044587410097394121/photo/1
中文: 需要 GPU,所以我决定尝试设置 EXO。 在本地网络上设置EXO,祝我好运 4090 + 4070 + 3070 + 2070 + Mac M1 Pro 32GB 让我们看看情况会如何,如果失败等等。 请告诉我要测试哪些模型 👇🏻
Dear algo: Buy me a GPU Thats all
中文: 亲爱的: 给我买一个GPU 全部
Guys if you want Gemma-4-E4B + Obilterated Trending on HF Front Page atm Check this out 👇🏻
中文: 如果你想要Gemma-4-E4B + Obilterated 高频首页趋势 请查看此处👇🏻
Today in AI NEWS — Slow day 💤 Here’s a recap: -Claude down (again) -New Claude Desktop app -OPUS 4.7 (soon) -SPUD (soon) -GPT 5.4 Cyber -Gemini CLI sub agents -Hermes-Workspace still on 🔥 -Need more VRAM (always) -Someone claimed they built Hermes Memory (false) 🏆OPEN…
中文: 今日在AI新闻中——慢日 💤 以下是一个回顾: - 再次放下 - 新的克劳德桌面应用程序 -OPUS 4.7(很快) SPUD(很快) -GPT 5.4 网络 - 双子座 CLI 子代理 - 爱马仕-工作空间仍在 🔥 - 需要更多VRAM(始终) 有人声称他们建立了爱马仕记忆(假) 🏆 已开启......
If you could build any tool/ software What would it be?? What’s stopping you from building it 👇🏻
中文: 如果你可以开发任何工具/软件 会是什么? 是什么阻碍了你建造它👇🏻
My X Premium + Subscription is paid Yet the platform has me stuck on Premium Asking support for help does absolutely nothing I swear 😂 Can ANYONE help me?? @Support @x @nikitabier @grok https://twitter.com/outsource_/status/2044505962631844019/photo/1
中文: 我的 X 高级订阅已付费 但这个平台让我一直依赖 Premium 请求帮助绝对没有任何帮助,我发誓😂 有人能帮我吗? @支持 @x @nikitabier @grok
Subagents are HERE in @GeminiCLI ⭐️ Spawn an entire team of specialized agents in parallel Gemini CLI just added subagents: 🧠Their own context windows + custom tools ✅Separate system prompts 🏆Run them simultaneously frontend-specialist etc. Delegate code reviews,…
中文: 子代理在 @GeminiCLI ⭐ 中 同时组建了一支专业特工团队 Gemini CLI 刚刚添加了子代理: 🧠 他们自己的上下文窗口 + 自定义工具 ✅ 分离系统提示 🏆 同时将它们处理为前端专家等。 委派代码审核,...
Community Noted 😂 https://twitter.com/outsource_/status/2044441075889185019/photo/1
中文: 社区已注😂
Squid Games but its with AI Agents 😂 Prompt 👇🏻 Squid Game-inspired AI frontier labs survival tournament. OpenAI GPT, Anthropic Claude, Google Gemini and Gemma, Qwen, MiniMax, DeepSeek, xAI Grok, Mistral, and Meta-style agents compete inside a giant futuristic arena. Sleek… https://twitter.com/outsource_/status/2044438729687757130/video/1
中文: 鱿鱼游戏,但采用人工智能代理😂 提示 👇🏻 受鱿鱼游戏启发的AI前沿实验室生存竞赛。OpenAI GPT、Anthropic Claude、Google Gemini 和 Gemma、Qwen、MiniMax、DeepSeek、xAI Grok、Mistral 和 Meta 风格的代理在一个巨大的未来主义竞技场中展开竞争。斯莱克......
RT @outsource_: 🚀 DFlash + MLX just dropped. What it does: ⚡️4x faster LLMs on Apple Silicon. 🚀Zero accuracy loss. 🔥Available right…
中文: RT @outsource_:🚀 DFlash + MLX 刚刚下线。 它做什么: ⚡️4x 更快的 LLM 在 Apple Silicon 上。 🚀 零精度损失。 [EE] 可说对......
At this point were averaging 100 ⭐️/ day Features in the works: -Office View for agents -Conductor (spec missions) -Operations (spec agents for automated tasks) + MORE! Drop any features 👇🏻 https://twitter.com/outsource_/status/2044433008132862289/photo/1
中文: 此时平均为100 ⭐️/ 作品中的特点: - 代理办公室视图 - 引值器(规格任务) - 操作(用于自动化任务的特点代理) +更多! 删除任何功能 👇🏻
Not Qwen 3.5 // Gemma etc thoooo
Twitter is 100x better when you connect with builders If you’re into: -Tech -AI -SaaS -Agents etc Say Hi 👋🏼
中文: 与构建者建立联系时,Twitter 的 100% 更好 如果你在: 科技 人工智能 SaaS 代理人 等 说你👋🏼
Builders and Early adopters of the HermesAgent movement Join the fasted growing X community 👇🏻 https://x.com/i/communities/2036289138844942498
中文: 赫尔梅斯特工运动的建造者和早期采用者 加入快速成长的X社区👇🏻
Can confirm guys I have Mythos Private access We have AGI
中文: 可以确认我有Mythos私人访问权的人 我们有AGI
RT @outsource_: NOW Hermes-Workspace Works w/ Atomic Chat 🔥 Shipped @atomic_chat_hq as a first-class provider: ✅ Auto-detect local mo…
中文: RT @outsource_:现在爱马仕-工作空间与原子聊天 🔥 以 @atomic_chat_hq 为一流供应商发货: ✅ 自动检测本地...
Tip of the day: Get yourself a pair of blue light blocking glasses. If your staring at the screen for hours a day Grab any pair from Amazon etc. 100% worth it! https://twitter.com/outsource_/status/2044420155933925761/photo/1
中文: 当天的提示: 给自己戴上一副蓝光遮挡眼镜。 如果你每天盯着屏幕看几个小时 从亚马逊等购买任何一对。 100% 值得!
This is what you need to do everyday Wake up with an agenda Feed it to your agent Have your agent take it from there. Whats stopping you?? Lack of experience? Lack of time?? https://twitter.com/outsource_/status/2044416643946717539/photo/1
中文: 这是你每天需要做的 带着议程醒来 把它喂给你的代理人 让你的经纪人从那里拿走。 有什么能阻止你? 缺乏经验?时间不够?
Can’t kill switch my local models tho 😉 Guys it’s more important to OWN hardware Dial in your local agent setup BUY A GPU Use and old gaming PC Hurry!!
中文: 无法杀死我的本地型号 😉 老家伙们,拥有硬件更重要 在您当地的代理机构中拨餐 购买 GPU 使用和旧游戏电脑 快点!
Who’s using OpenClaw / HermesAgent To TRADE daily? Who uses it for poly market? Is this an actual use case of just posted for clicks??
中文: 谁在使用 OpenClaw / HermesAgent 每天要交易? 谁用它来做聚酯市场? 这只是一个刚刚发布来点击的实际使用情况吗?
Hot Take: Handoff's from session to session is🔑 Helps you 10x output and context with your agent! If you aren't using your agents to streamline PRS You're NGMI TIP👉🏻 Remember always set a goal for EACH session! Don't expect to complete an entire project in one. https://twitter.com/outsource_/status/2044285049302839717/photo/1
中文: 热门内容:从会话到会话的 Handoff 是🔑 帮助您与代理进行10次输出和上下文! 如果你没有使用你的代理来简化公关 你是NGMI 提示👉🏻 请记住,每个环节总会设定目标! 不要指望完成一个完整的项目。
NOW Hermes-Workspace Works w/ Atomic Chat 🔥 Shipped @atomic_chat_hq as a first-class provider: ✅ Auto-detect local models on startup 🔌 Zero config just open Workspace, models sync 💬 Chat routes directly to your local server ⚡ Powered by TurboQUANT via Atomic Chat +… https://twitter.com/outsource_/status/2044283120023748864/photo/1
中文: 现在 Hermes-Workspace 与 Atomic Chat 合作 🔥 以 @atomic_chat_hq 为一流供应商发货: ✅ 在启动时自动检测本地模型 🔌 零配置只需打开工作区,模型同步 💬 直接与本地服务器进行聊天 ⚡ 由 TurboQUANT 通过 Atomic Chat 提供支持
🚀 DFlash + MLX just dropped. What it does: ⚡️4x faster LLMs on Apple Silicon. 🚀Zero accuracy loss. 🔥Available right now. How it works: 🤯Tiny block diffusion model drafts tokens in parallel. 🚨Main LLM verifies in one shot. Try it now 👇🏻
中文: 🚀 DFlash + MLX 刚刚下线。 它做什么: ⚡️4x 更快的 LLM 在 Apple Silicon 上。 🚀 零精度损失。 [EE] 可立即批准。 它是如何工作的: 🤯 微小的区块扩散模型并行地起草代币。 🚨 主传球通过一击验证。 立即试用👇🏻
What do we like better for Hermes-Workspace Mobile? Light/ Dark Mode 🌙 https://twitter.com/outsource_/status/2044258273407176717/photo/1
中文: 我们更喜欢 Hermes-Workspace Mobile 的什么? 浅色/暗色模式 🌙
Hermes-Workspace growth has been INSANE 🔥 Over 10,000+ Clones overall 30,000+ Views overall 5000+ Unique Cloners overall 1400+ GitHub⭐️ https://twitter.com/outsource_/status/2044231349263806922/photo/1
中文: 爱马仕与办公空间的增长一直是INSANE 🔥 超过10,000个克隆 超过30,000次观看 5000 多个独特克隆体 1400+ GitHub⭐️
Shoutout to everyone who's contributed to Hermes-Workspace ⭐️ New features dropping soon! Reply any bugs or new features 👇🏻 https://twitter.com/outsource_/status/2044230511023870414/photo/1
中文: 向所有做出贡献的人大放不週 爱马仕-工作空间 ⭐️ 新功能即将推出! 回复任何漏洞或新功能 👇🏻
RT @outsource_: Guys 🤯 SuperGemma 4 31B Abliterated DROPPED By popular demand @songjunkr released 🔥 🚨Built on Gemma 4 31b-it-base and opt…
中文: RT @outsource_: Guys 🤯 SuperGemma 4 31B 已被删除 热门点播 @songjunkr 发布 🔥 🚨 基于 Gemma 4 31b-it 基础,选择......
When should I drop the DISCORD Chat ?? You can chat with my Agent inside.. Will have FREE Guides For Beginners in AI + More 🚀 Inside: 💬 COMMUNITY 📚 LEARNING 🛠️ TOOLS & STACKS 🎙️ VOICE +💎 MENTORSHIP (for those interested) 100% FREE. Join. Learn. Thats it 🤯 https://twitter.com/outsource_/status/2044197274767503640/photo/1
中文: 何时应该放弃《DISCORD》聊天? 你可以和我的经纪人聊天。 将为人工智能及更多初学者提供免费指南 🚀 内部: 💬 社区 📚 学习 🛠️ 工具与放大器;STACKS 🎙️ 声音 +💎 导师关系(适合感兴趣的人) 100%免费。加入。学习。就是这样🤯
When should I drop the DISCORD Chat ?? The best place to kickstart your journey with agents Will have FREE Guides For Beginners in AI + More 🚀 Inside: 💬 COMMUNITY 📚 LEARNING 🛠️ TOOLS & STACKS 🎙️ VOICE +💎 MENTORSHIP (for those interested) 100% FREE. Join. Learn. Thats it… https://twitter.com/outsource_/status/2044196304398491773/photo/1
中文: 何时应该放弃《DISCORD》聊天? 与经纪人一起开启旅程的最佳地点 将为人工智能及更多初学者提供免费指南 🚀 内部: 💬 社区 📚 学习 🛠️ 工具与放大器;STACKS 🎙️ 声音 +💎 导师关系(适合感兴趣的人) 100%免费。加入。学习。就是这样......
Guys 🤯 SuperGemma 4 31B Abliterated DROPPED By popular demand @songjunkr released 🔥 🚨Built on Gemma 4 31b-it-base and optimized: 🤯 Uncensored (no safety lobotomy) ⚙️Lightweight (4-bit quantized) 🧠Actually smart (smokes the base model) Benchmark Gains (vs 31b-it-base)… https://twitter.com/outsource_/status/2044189743865287120/photo/1
中文: 球员🤯 超级宝石 4 31B 被淘汰 热门点播 @songjunkr 发布 🔥 🚨 基于 Gemma 4 31b 基座并优化: 🤯 未经审查(无安全定位) ⚙️ 轻量化(4位量化) 🧠 智能(模拟基础模型) 基准收益(与31b-it-base)...
There needs to be a foundation for Open Source devs Literally shipping models left and right Building on what frontier labs ship Adding their own spice Distillation + Abliteration ETC What a time to be in AI.
中文: 开源开发者需要有基础 字面上运输的型号左右 以前沿实验室的船只为基础 添加自己的香料 蒸馏 + 消蒸电子化 人工智能中应该有多好的时间。
Bro you can’t make this up… One lab pushes for something The other lab does the same. No hate but can we be more original?? Obviously this is a very important issue But I’d like to see labs pushing NEW things It’s literally monkey see… money do 😂
中文: 兄弟,你无法编得一闘...... 一个实验室推动某件事 另一个实验室也做了同样的事情。 没有仇恨,但我们又能更原始吗? 显然这是一个非常重要的问题 但我希望看到实验室推动新事物 简直是猴子......钱真的有......
If you’re new here say Hi 👋🏼 Reply for a follow back 👇🏻 https://twitter.com/outsource_/status/2044170529939013993/photo/1
中文: 如果你是新来的,请说你好👋🏼 回复回复 👇
RT @outsource_: 🚨QWEN 3.5 40B DENSE + FULL HERETIC + OPUS 4.6 🤗@huggingface model card: 🧠 40 Billion dense parameters (NOT MoE) 📈 Expand…
中文: RT @outsource_:🚨QWEN 3.5 40B 密度 + 完整 HERETIC + OPUS 4.6 🤗@huggingface 模型卡片: 🧠 400亿个密集参数(非MOE) 📈 扩展...
🚨QWEN 3.5 40B DENSE + FULL HERETIC + OPUS 4.6 🤗@huggingface model card: 🧠 40 Billion dense parameters (NOT MoE) 📈 Expanded from 27B → 96 layers + 1,275 tensors 🧪 Multi-stage trained on Claude 4.6 Opus High ⚔️ Fully Heretic-Uncensored first (abliterated) 🔧 Upgraded… https://twitter.com/outsource_/status/2044154525909467472/photo/1
中文: 🚨 QWEN 3.5 40B 密度 + 完整 HERETIC+ OPUS 4.6 🤗@huggingface 模型卡片: 🧠 400亿个密集参数(非MOE) 📈 扩展至27B → 96层 + 1,275 个张力 🧪 在克劳德4.6 Opus High上进行了多级训练 ⚔ 完全异端无删素(已删除) 🔧 升级...
MiniMax just updated the license by popular request! This is a HUGE win for OpenSource The future will be OpenSource 100% Buy GPUS now!!!!
中文: miniMax 刚刚根据热门要求更新了许可证! 这对开源而言是一个巨大的胜利 未来将是开源的100% 立即购买 GPU!!
Wait?? Is this??? CODEX 😂😂 https://twitter.com/outsource_/status/2044133552514564421/photo/1
中文: 等等??这是吗??CODEX 😂😂
🚨 Launching Agent Powered Discord for Builders & Beginners… New to AI agents / local models + OpenClaw/ HermesAgent? My agent Aurora is setting it up while I type this.. 🚀 Jump in and chat with her directly for help, debugging, or roadblocks. https://twitter.com/outsource_/status/2044132916775493799/photo/1
中文: 🚨 推出适用于 Builders &amp 的 Agent Powered Discord;初学者...... 向人工智能代理/本地模型提供新功能 + OpenClaw/HermesAgent? 我的经纪人Aurora正在设置它,而我输入此内容。🚀 直接跳进去与她聊天,寻求帮助、调试或设置障碍。
Are you a Founder? Developer? Designer? User?
中文: 你是创始人吗? 开发者? 设计师? 用户?
great product @tailscale 🔥
中文: 出色产品 @tailscale 🔥
Good Feature but Id rather own my own infrastructure Antrhopic wants you to run all your agents through them The future is OPEN not CLOSED You will run whatever agent you want + harness If you need a lab to do this for you You're NGMI
中文: 功能好,但我更想拥有自己的基础设施 Antrhopic希望你通过他们来管理所有代理 未来尚未结束 你将运行任何你想要的代理 + 线束 如果你需要一个实验室来为你做这件事 你是NGMI
RT @songjunkr: SuperGemma4와 같은 로컬모델들은 세팅에 매우 민감합니다. SuperGemma-fast : Text only, 140+tok/s, 매우 빠른모델 SuperGemma-Multimodal : 비전 및 에이전트 툴콜링…
What happened to prompt engineers I think they got replaced by Common sense?? LOL
中文: 工程师们发生了什么事 我觉得他们被取代了 常识? 哈哈
I’m gonna be dropping FREE guides On how to get started with AI Guys it’s not too late. We’re in a bubble. Go ALL IN on AI / AGENTS Ask Chat GPT + Grok Use any OLD PCS It’ll be worth it trust me 🙏🏼
中文: 我要放弃免费指南了 如何开始使用AI 老是不晚了。 我们正处于一个泡沫之中。 全力以赴,了解人工智能/代理 询问聊天 GPT + Grok 使用任何旧电脑 相信我是值得的🙏🏼
Find new trending models Post them Repeat 🔁 200k+ views in less than 24HRS And I just crossed 4k followers Let’s gooo 🚀🚀🚀 https://twitter.com/outsource_/status/2044047654280212520/photo/1
中文: 寻找新的趋势模型 发布它们 重复🔁 2亿+次观看次数少于24小时 我刚刚跨越了4k个粉丝 我们来说🚀🚀[EE]
When I post models they go viral guys
中文: 当我发布模特时,他们会走红
We’ve reached the singularity @TheAhmadOsman uses Hermes-Workspace 🤯🤯
中文: 我们已经达到了@TheAhmadOsman使用Hermes-Workspace的奇点性🤯🤯
We’ve reached the singularity @TheAhmadOsman uses Hermes-Workspace 🤯🤯
中文: 我们已经达到了@TheAhmadOsman 使用的奇点 爱马仕-工作空间 🤯🤯
I can notice more people pushing long form content but those videos seem to have the same if not worse reach than actual viral style posts. It seems like the algo was correct. They want users to post videos to train their models. If you see long form videos being pushed please…
中文: 我可能会注意到,更多人在推送长篇内容,但这些视频似乎与实际的流行风格帖子相比,其影响力似乎同样,即使不差。 看来那是对的。他们希望用户发布视频来训练自己的模型。 如果看到长视频被推送,请......
For those wanting tool calling etc for HermesAgent / OpenClaw run this https://huggingface.co/Jiunsong/supergemma4-26b-abliterated-multimodal
中文: 对于那些希望为 HermesAgent / OpenClaw 提供工具调用等的人,请使用此信息
If you want to benchmark local models on your hardware Use this tool Local Mac app & more just shipped Working on a community aspect of this and other bench-loop style things.
中文: 如果你想在硬件上对本地模型进行基准测试 使用此工具 本地 Mac 应用和应用;更多只是发货 从事此项和其他单板循环模式的社区层面工作。
RT @outsource_: If you’re NEW in the AI Space And you need HELP setting up: Agents/ Local Models / Harnesses (OpenClaw or HermesAgent etc…
中文: RT @outsource_:如果你在人工智能领域是新人 您需要帮助设置: 代理/本地模型/琴饰(OpenClaw 或 HermesAgent 等)
Stripe/ Whop/ all payment processors right now https://twitter.com/outsource_/status/2043892881836892413/photo/1
中文: Stripe/ Whop/ 所有支付处理软件,目前
Vote and share guys 👇🏻 https://twitter.com/outsource_/status/2043874842223477046/photo/1
中文: 投票并分享人员 👇🏻
Need more compute
中文: 需要更多计算
If you’re NEW in the AI Space And you need HELP setting up: Agents/ Local Models / Harnesses (OpenClaw or HermesAgent etc) I want to help YOU 🫵🏻 100% Free. No strings attached. My only questions is how 👇🏻
中文: 如果你在人工智能领域是新人 您需要帮助设置: 代理/本地模型/连接(OpenClaw 或 HermesAgent 等) 我想帮助你🫵🏻 100%免费。没有附加条件。 我唯一的问题是如何👇🏻
How I got 5M+ impressions in 2 weeks on X 👇🏻 I didn't learn how to code. I learned how to use agents. From 0 coding experience -> shipping 2 products That crossed 1,600+ GitHub ⭐️ The real skill now is not writing every line yourself. It’s knowing what to build, how to… https://twitter.com/outsource_/status/2043848791053258900/photo/1
中文: 如何在X上获得5M+的印象 我没有学会编程。 我学会了如何使用代理。 从0个编码经验 ->发货2种产品 超过1600多个 GitHub ⭐️ 现在真正的技能并不是自己写下每一行。 它知道要构建什么,如何......
X is literally a FULL TIME job You have to be locked in Reply Engage Post Differentiate Whilst living your life For free? 😂 Hopefully this first payout is worth my time
中文: X 简直就是一份完整时间的工作 你必须被锁在 回复 订婚 发布 差异化 在过你的生活 免费吗?😂 希望这次首次付款值得我花时间
Yo @nikitabier @X @grok Please fix. I have X Premium+ It renewed today And I’m seeing non stop ads. Problem with the system in place Can’t stand ads. lol. Waiting on support https://twitter.com/outsource_/status/2043841427348558267/photo/1
中文: @nikitabier @X @grok 请修好。我有 X Premium+ 今天更新了 我看到的是不间断的广告。 系统存在问题 无法忍受广告。等待支持
What do you guys think? Is this the result of them throttling OPUS? So by default its boosted? Nerfing their model to sell the new one sounds like them Obviously Mythos is probably very capable. But users reporting degraded performance And this launch isnt a coincidence. SAME…
中文: 你们觉得怎么样? 这是他们限制OPUS的结果吗? 那么默认情况下,它的提升是什么? 让他们的模特卖新款听起来很像 显然,神话可能非常有能力。 但用户报告性能下降 这次发布并非巧合。 同样的......
RT @outsource_: 🚨SUPER GEMMA 4 26B UNCENSORED MLM LANDED @songjunkr dropped SuperGemma4-26B-Abliterated-Multimodal-MLX-4bit The perfect…
中文: RT @outsource_:🚨 超级 GEMMA 4 26B 未受CENSORED MLM LANDED @songjunkr 放弃了 SuperGemma4-26B-Abliteed-Multimodal-MLX-4bit 完美的......
RT @outsource_: Hermes Agent Fork Updated to v0.9.0🔧 Merged the latest upstream release into our fork while keeping the frontend session A…
中文: RT @outsource_:爱马仕代理分叉更新至 v0.9.0🔧 将最新的上游版本合并到我们的分叉中,同时保持前端会话A...
RT @outsource_: https://hermes-workspace.com/ Chat Feature ⭐️ Guys if you didn't know CMD + F for a built in SEARCH If you didn't know, now you…
中文: RT @outsource_: 聊天功能 ⭐️ 如果你不了解CMD + F,适合在搜索中构建 如果你不知道,现在你......
🚨SUPER GEMMA 4 26B UNCENSORED MLM LANDED @songjunkr dropped SuperGemma4-26B-Abliterated-Multimodal-MLX-4bit The perfect lighter-weight beast for every Mac user! This thing SMOKES the regular Gemma-4 26B : 🤯0/100 refusals — fully abliterated & uncensored 😉Full… https://twitter.com/outsource_/status/2043817028176253227/photo/1
中文: 超级GEMMA 4 26B未加查的真融会 @songjunkr 放弃了 SuperGemma4-26B-Abliteed-Multimodal-MLX-4bit 每位Mac用户都完美的轻量级野兽! 这款软件可与常规的 Gemma-4 26B 进行 smok: 🤯/100 拒绝——完全取消和删减;未经审查 😉 完整...
One of the many features built into HermesWorkspace
中文: HermesWorkspace 内置的众多功能之一
https://hermes-workspace.com/ Chat Feature ⭐️ Guys if you didn't know CMD + F for a built in SEARCH If you didn't know, now you know I have so many features packed in I forget LOL Try it today 👇🏻 https://twitter.com/outsource_/status/2043809133871214961/photo/1
中文: 聊天功能 ⭐️ 如果你不了解CMD + F,适合在搜索中构建 如果你不知道,现在就知道了 我有这么多功能,都忘了LOL 立即尝试👇🏻
Finding out someone new blocked me Just amazes me What do you not like about what I say? I’m very based Straight to the point 😂 https://twitter.com/outsource_/status/2043799968113164352/photo/1
中文: 发现一个新人挡住了我 只是让我惊叹 你不喜欢我说什么吗? 我非常有基础 直接到点 😂
LOL
Hermes Agent Fork Updated to v0.9.0🔧 Merged the latest upstream release into our fork while keeping the frontend session API patch in place for Hermes Workspace. ⭐️Merged upstream v0.9.0 -Pulled in the new local web dashboard -Added /fast support -Brought in iMessage +… https://twitter.com/outsource_/status/2043792292889215049/photo/1
中文: 爱马仕代理叉更新至v0.9.0🔧 将最新的上游版本合并到我们的工作库中,同时为 Hermes Workspace 保留了前端会话 API 补丁。 ⭐️合并上游 v0.9.0 在新的本地网页仪表板中 - 已添加/快速支持 以iMessage+获取......
So questions guys... Do you think these Frontier labs strictly use their own models? When developing things do they swap out for others like Claude Opus // GPT 5.4? Or do they strictly use their own? What do you think
中文: 所以,有人在提问...... 你认为这些前沿实验室是否严格使用自己的模型? 发展事物时,是否会换成别人 克劳德·奥普斯 / / GPT 5.4? 还是他们严格使用自己的?你觉得怎么样
Guys I built my agent Aurora since February 2026⚡️ It has been running: 11,000+ Messages 6,000+ Commits 5,000+ Tool Calls 73 + Days 43M Tokens ASK ME ANYTHING 👇🏻 https://twitter.com/outsource_/status/2043787397821874472/photo/1
中文: 自2026年2月以来,我建立了我的经纪人Aurora⚡️ 一直在运行: 11000 多条消息 6000多份提交 5000 多个工具调用 73天 43M代币 请向我发送任何信息 👇🏻
Are you guys having reliability issues with your Agents? What are you running? HermesAgent? OpenClaw? Other? Reply with any issues ill help 👇🏻 https://twitter.com/outsource_/status/2043786850339299538/photo/1
中文: 你们的经纪人有可靠性问题吗? 你在跑步什么? HermesAgent?OpenClaw? 其他? 回复任何问题,帮助 👇🏻
Guys use frontier models + agentic harness's To get YOUR Agent as capable as possible. It does matter what Harness you use (use the most capable one) HermesAgent It will shortcut you the trial and error of others (OpenClaw) Get your agent dialed in Then swap out with local… https://twitter.com/outsource_/status/2043784911589392452/photo/1
中文: 男士们使用前沿模型 + 特质线束 让您的经纪人尽可能有能力。 使用什么手法很重要 (使用最有能力的)HermesAgent 它会让你在别人的反复尝试和犯错中走捷径 (OpenClaw) 让你的代理人被拨入 然后与本地人交换......
Shoutout to @huggingface 🤗 Hands Down the BEST place to find: Trending Open Models Uncensored Models MLX Models GGUF Models TurboQuant models ETC. If you aren’t using them yet, go check them out 🚨 https://twitter.com/outsource_/status/2043784650477240521/photo/1
中文: 向@huggingface大声喊叫 🤗 放下最佳位置: 趋势开放模型 无审查模型 MLX 模型 GGUF 模型 TurboQuant 车型 ETC。 如果尚未使用,请前往查看 🚨
GUYS ITS OUT NOW GO UPDATE YOUR HERMES AGENT UPDATE NOW! https://twitter.com/outsource_/status/2043775914928984504/photo/1
中文: 你现在的出局会更新你的爱马仕 立即更新!
BUY A GPU BUY A GPU BUY A GPU BUY A GPU BUY A GPU
Take a look at the agent that shipped @ClawSuite & https://hermes-workspace.com/ Drop Aurora a ❤️
中文: 请查看发货的代理 @ClawSuite & 将极光滴下❤️
Alright who’s got access to Claude Mythos Preview on Bedrock?? Send it to me, I’ll pay $1000 per run 😂 https://twitter.com/outsource_/status/2043766762311692657/photo/1
中文: 谁可以访问《卧底》上的克劳德·弗利多斯预览版? 给我,我每次跑步支付1000美元😂
Good work on this @stevibe Ill be running this and seeing what I can extract / contribute Im in the middle of building a BenchMark Loop system for local models similar to this. We can merge the concepts if it fits, ill let you know! https://twitter.com/outsource_/status/2043763804555657528/photo/1
中文: @stevibe 的优秀工作 负责运行此内容并查看我能提取/贡献的内容 正在为类似本地模型构建BenchMark循环系统。 如果符合,我们可以将概念合并,但不要告诉您!
Take a look at the agent that shipped @clawusite & https://hermes-workspace.com/ Drop Aurora a ❤️
中文: 请查看发货的代理 @clawusite & 将极光滴下❤️
Guys very important if you are using OpenClaw DONT Update 😂 you'll probably break your setup If running HermesAgent 100% UPDATE they ship usable features You're welcome - MGMT
中文: 使用OpenClaw非常重要的人 不要更新😂,你可能会破坏你的设置 如果运行 HermesAgent 100% 更新,它们会提供可用的功能 欢迎光临 - MGMT
Guys Opus 4.6 IS NERFED It’s still good, personality is on point But it feels less autonomous Couldn’t be placebo Let me know your experience 👇🏻
中文: 男生 Opus 4.6 已 已 中和 仍然很好,个性在点上 但感觉不太自主 不能是安慰剂 请告诉我您的经验 👇🏻
WE DID IT GUYS 🤯
中文: 我们确实对它🤯
Yooo @nikitabier start sending those checks 👋🏻 WE DID IT CHAT 👋🏻 Lets gooo Build In Public guys trust! Reply with any questions you have 👇🏻 https://twitter.com/outsource_/status/2043736749759709629/photo/1
中文: 尤 @nikitabier 开始寄送这些支票 👋🏻 我们进行了聊天👋🏻 让你好 在公众中建立,人们信任! 有任何问题,请回答👇🏻
RT @outsource_: Go submit your agent @DiscoverHermes See how it ranks! Im #3 🚀 https://twitter.com/outsource_/status/2043543760563253386/photo/1
中文: RT @outsource_:去提交你的代理 @DiscoverHermes 看看它是如何排列的!网址:第3名 🚀
RT @outsource_: 🚨 NEW MiniMax updates the M2.7 license after massive backlash! @RyanLeeMiniMax clarified: ✅ Self-hosted for personal c…
中文: RT @outsource_:🚨 全新MiniMax在遭到强烈抵制后更新了M2.7许可证! @RyanLeeMiniMax 澄清道: ✅ 个人托管自托管......
RT @outsource_: 🚨 SUPER GEMMA 4 26B UNCENSORED IS INSANE LLM WIZARD COOKING AGAIN @songjunkr Dropped SuperGemma4-26B-Uncensored GGUF v2 a…
中文: RT @outsource_:🚨 超级 GEMMA 4 26B 未检查是 INSANE LLM WIZARD 再次烹饪 @songjunkr 放弃 SuperGemma4-26B-无审查 GGUF v2 a...
🚨 SUPER GEMMA 4 26B UNCENSORED IS INSANE LLM WIZARD COOKING AGAIN @songjunkr Dropped SuperGemma4-26B-Uncensored GGUF v2 and it’s trending on @huggingface🤗 This thing SMOKES the regular Gemma-4 26B: 🤯0/100 refusals (actually uncensored) 🚀Fixed all the tool-call +… https://twitter.com/outsource_/status/2043710372176122085/photo/1
中文: 🚨 超级GEMMA 4 26B 未受审查是无 LLM WIZARD 再次烹饪 @songjunkr 放弃 SuperGemma4-26B-Uncensored GGUF v2,其趋势在 @huggingface🤗 上 这款内容可与常规的 Gemma-4 26B 进行脱节: [EE2/0/100 拒绝(实际上未经审查) 🚀 修复了所有工具调用 +...
🚨 SUPER GEMMA 4 26B UNCENSORED IS INSANE LLM WIZARDWIZARD COOKING AGAIN @songjunkr Dropped SuperGemma4-26B-Uncensored GGUF v2 and it’s trending on @huggingface🤗 This thing SMOKES the regular Gemma-4 26B: 🤯0/100 refusals (actually uncensored) 🚀Fixed all the tool-call +… https://twitter.com/outsource_/status/2043709922857144822/photo/1
中文: 🚨 超级GEMMA 4 26B 未受审查是无 LLM WIZARDWIZARD 再次烹饪 @songjunkr 放弃 SuperGemma4-26B-Uncensored GGUF v2,其趋势在 @huggingface🤗 上 这款内容可与常规的 Gemma-4 26B 进行脱节: [EE2/0/100 拒绝(实际上未经审查) 🚀 修复所有工具调用 +...
Guys it’s more important NOW more than ever to host your agent locally. You need to own your agent. Don’t host it on someone’s server I agree you should power it by cloud models (to a point) But locally hosted not on a VPS + Strong cloud model You can use a local model as…
中文: 现在比以往任何时候都更需要在本地托管你的代理。 你需要拥有你的代理人。 不要将其托管在某人的服务器上 我同意你应该通过云模型来推动它(在某种程度上) 但本地托管的不是VPS +强云模型 可以使用本地模型作为...
🚨 NEW MiniMax updates the M2.7 license after massive backlash! @RyanLeeMiniMax clarified: ✅ Self-hosted for personal coding & internal workflows = FREE & allowed ✅ Distillation fully supported for personal use, internal tools & non-profits ❌ Large-scale commercial…
中文: 🚨 NEW MiniMax 在遭到强烈抵制后更新了 M2.7 许可证! @RyanLeeMiniMax 澄清道: ✅ 个人编程自助式;内部工作流程 = 免费&允许 ✅ 完全支持个人使用、内部工具和非营利组织 ❌ 大规模商业...
Go submit your agent @DiscoverHermes See how it ranks! Im #3 🚀 https://twitter.com/outsource_/status/2043543760563253386/photo/1
中文: 提交您的代理 @DiscoverHermes 看看它是如何排列的!网址:第3名 🚀
I shared my agent on @DiscoverHermes Aurora ⚡️ dual-platform AI copilot that ships open-source agent tools with a solo founder (me) https://discoverhermes.com/use-cases/24 https://twitter.com/outsource_/status/2043533843953496067/photo/1
中文: 我在@DiscoverHermes上分享了我的经纪人 Aurora ⚡️ 与独自创始人一起发布开源代理工具的双平台人工智能副驾驶
RT @outsource_: Adding this to Hermes-Workspace 🚀 https://twitter.com/outsource_/status/2043370541776576752/photo/1
中文: RT @outsource_:将此内容添加到 Hermes-Workspace 🚀
What a company...
中文: 多么一家公司......
What a great company @anthropicai is @claudeai loves their users 😂 NOT literally horrible community. Shit PR Smh https://twitter.com/outsource_/status/2043484036694225033/photo/1
中文: @sclaudeai 是 @claudeai 喜爱用户的好公司 😂 并非真正糟糕的社区。 清真公关 请访问
New HermesWorkspace update launched How to update 👇🏻
中文: 新HermesWorkspace更新启动 如何更新 👇🏻
RT @outsource_: HERMES-WORKSAPCE IS CROSSING 1,200+ ⭐️ STARS LETS GOOO 🚀 Shoutout to all the community support and contributors on the…
中文: RT @outsource_:HERMES-WORKSAPCE 正在突破 1200 多星 LETS GOOO 🚀 向所有社区支持和贡献者大喊大叫......
RT @outsource_: 🚨NEW Release for Hermes-Workspace Just Landed + Hermes Agent 🔧 Whats Inside 👇🏻 NEW -> Task Board / Kanban Page NEW -> Ag…
中文: RT @outsource_:🚨 为 Hermes-Workspace 发布 + 爱马仕经纪人 🔧 内在的什么👇🏻 新 -> 任务板 / 看板 新 -> 农业...
HERMES-WORKSAPCE IS CROSSING 1,200+ ⭐️ STARS LETS GOOO 🚀 Shoutout to all the community support and contributors on the latest release! Hermes Workspace is still taking off Let me know any features/ issues 👇🏻 https://twitter.com/outsource_/status/2043444161810997251/photo/1
中文: 赫默斯-沃克斯普斯将拥有超过1200颗星星 LETS GOOO 🚀 向最新发布的所有社区支持和贡献者大放不清! 爱马仕工作区仍在起飞 请告诉我任何功能/问题 👇🏻
HERMES-WORKSAPCE CROSSED 1,200 ⭐️ STARS LETS GOOO 🚀 Shoutout to all the community support and contributors on the latest release! Hermes Workspace is still taking off Let me know any features/ issues 👇🏻 https://twitter.com/outsource_/status/2043444035180765260/photo/1
中文: 赫默斯-沃克斯普斯穿越了1200颗⭐️星 LETS GOOO 🚀 向最新发布的所有社区支持和贡献者大放不清! 爱马仕工作区仍在起飞 请告诉我任何功能/问题 👇🏻
HERMES-WORKSAPCE CROSSED 1,200 ⭐️ STARS LETS GOOO 🚀 Shoutout to all the community support and contributors on the latest release! Hermes Workspace is still taking off Let me know any features/ issues 👇🏻 https://twitter.com/outsource_/status/2043443997188735257/photo/1
中文: 赫默斯-沃克斯佩斯穿越了1200颗⭐️星 LETS GOOO 🚀 向最新发布的所有社区支持和贡献者大声呐喊! 爱马仕工作区仍在起飞 请告诉我任何功能/问题 👇🏻
🚨NEW Release for Hermes-Workspace Just Landed + Hermes Agent 🔧 Whats Inside 👇🏻 NEW -> Task Board / Kanban Page NEW -> Agent Artifacts in Inspector NEW -> Crew Monitoring for Agent Profiles NEW -> Multi-Language Support NEW -> Configurable Knowledge Base Sources NEW -> Skills… https://twitter.com/outsource_/status/2043438707500368220/photo/1
中文: 爱马仕-工作空间刚刚登陆的最新发布 + 爱马仕经纪人 🔧 内在的内容 👇🏻 新 -> 任务板 / 看板 新 -> 探员代理文物 新 -> 代理配置文件的机组监控 全新 -> 多语言支持 全新 -> 可配置的知识库 新技能......
RT @outsource_: It’s open source guys but… You can’t use it to ship anything for commercial purposes I get it, but why put limitations on…
中文: RT @outsource_:这是开源的人,但是...... 无法将其用于商业目的的发货 我了解它,但为什么要对......施加限制呢
The last thing Claude Opus sees before Dario NERFS it https://twitter.com/outsource_/status/2043393769995747405/photo/1
中文: 克劳德·奥弗斯在《达里奥·内尔斯》之前最不见容的事
RT @outsource_: Guys you can run MiniMax 2.7 with 60GB VRAM Use this quant from JANGQ AI and post results!
中文: RT @outsource_:您可以使用60GB VRAM运行MiniMax 2.7的人 使用 JANGQ AI 的此量子 并发布结果!
Adding this to Hermes-Workspace 🚀 https://twitter.com/outsource_/status/2043370541776576752/photo/1
中文: 将此内容添加到 Hermes-Workspace 🚀
Guys you can run MiniMax 2.7 with 60GB VRAM Use this quant from JANGQ AI and post results!
中文: 您可以使用60GB VRAM运行MiniMax 2.7 使用 JANGQ AI 的此量子 并发布结果!
🚨 @UnslothAI Quants Dropped For MiniMax 2.7 🔥 Unsloth just dropped day-zero optimized support for the new open-source MiniMax-M2.7 So you can run this 230B MoE monster at home. ⚡ HOW UNSLOTH CRUSHED THE QUANTIZATION - Applied their brand-new Unsloth Dynamic 2.0 tech - Used… https://twitter.com/outsource_/status/2043344011277066563/photo/1
中文: @UnslothAI Quants 已下注至 MiniMax 2.7 🔥 Unsloth刚刚放弃了对全新开源MiniMax-M2.7的零日优化支持 所以你可以在家里运行这个230B MoE怪物。 ⚡ 不合群如何粉碎了量化 - 应用他们全新的 Unsloth Dynamic 2.0 技术 - 已使用......
100,000 Impressions Away from being monetized on @X It’s taken me 2 weeks of being locked in Learning the algo And being consistent Some days I take it slow Some days I grind more. Show up everyday and be consistent 🚀 https://twitter.com/outsource_/status/2043341323898404889/photo/1
中文: 10万 印象远不止在 @X 上被变现 被关在了两周前 学习吧 保持一致 有些日子我慢慢来 有些日子我磨得更多。 每天展示并保持一致🚀
RT @Govindtwtt: They lied to you. Building software is really hard.
中文: RT @Govindtwtt:他们对你撒了谎。构建软件真的很难。
DON’T USE BREAKING IN YOUR HEADLINE 😂 Don’t use breaking anymore guys we’re cooked. They are going hard to disrupt payouts Targeting pages that use that Or post unoriginal context 100x per day Will learn the new algo. Test different styles of content and see what happens.
中文: 不要在标题中使用 breaking 😂 别再用我们做饭的家伙了。 他们很难扰乱支出 针对使用该内容的页面 或每天发布非原创背景内容100次 将学习新的阿戈。 测试不同类型的内容,并了解情况如何。
BIG NEWS: X will prioritize ORIGINAL content sources for payouts Gives users more incentives to actually publish original content instead of just copying headlines for views! Not sure the exact way this is tracked, probably using grok. But it’s getting gray area if they are…
中文: 重大新闻:X将优先关注原始内容来源以获取赔付 为用户提供更多激励,让用户能够真正发布原创内容,而不仅仅是将标题复制为观看内容! 不确定具体追踪方式,可能使用grok。但如果是......,情况就会变得灰暗
It’s open source guys but… You can’t use it to ship anything for commercial purposes I get it, but why put limitations on an opensource model? Doesn’t that defeat the purpose
中文: 是开源的,但是...... 无法将其用于商业目的的发货 我了解它,但为什么要对开源模型设置限制呢? 难道这不能破坏目的吗
RT @outsource_: BREAKING: 🤯 MINIMAX JUST OPEN-SOURCED M2.7 THE SELF-EVOLVING AGENTIC BEAST IS OPEN 🔥 ⭐️229B params. 🔥Native Agent Teams…
中文: RT @outsource_: 突发:🤯 MINIMAX 只是开源的 M2.7 自我突破的代理野兽是开放的🔥 ⭐️229B 参数。 🔥 原发性特工团队......
BREAKING: 🤯 MINIMAX JUST OPEN-SOURCED M2.7 THE SELF-EVOLVING AGENTIC BEAST IS OPEN 🔥 ⭐️229B params. 🔥Native Agent Teams + dynamic tool use + 🌐autonomous optimization loops. 🪄SOTA drops: 56.22% SWE-Pro 🚀57.0% Terminal Bench 2 🚨GDPval-AA ELO 1495 (highest open model)… https://twitter.com/outsource_/status/2043140342455648390/photo/1
中文: 突破:MINIMAX 刚刚开源 M2.7 自我突破的代理野兽是开放的🔥 ⭐️229B 参数。 🔥 原生代理团队 + 动态工具使用 + 🌐 自主优化循环。 🪄 下降:56.22% SWE-Pro 🚀57.0% 航站楼二号 🚨 GDPD级-AA ELO 1495(最高开放模式)......
Wyd? Nothing much Just orchestrating agent maxxing Claude Opus Maxxing Codex Maxxing What are YOU doing 🫵🏻 https://twitter.com/outsource_/status/2043132851302859111/photo/1
中文: 是吗? 没什么 只需编排代理maxxing 克劳德·奥弗斯·马克斯兴 马克兴法典 你正在做什么🫵🏻
What’s the product or tool you wish was built?? What’s stopping you from shipping it Reply 👇🏻👇🏻
中文: 你希望生产什么产品或工具? 是什么阻碍了你发货 回复 👇🏻👇
RT @Web3_Matters: 1/ Most people are asking the wrong question about AI agents. It’s not “Openclaw or Hermes?” It’s: Can people actually…
中文: RT @Web3_Matters: 1/ 大多数人对人工智能代理提出了错误的问题。 不是“Openclaw 还是 Hermes?” 是: 人们真的能......
RT @outsource_: Guys it’s 3 AM Are you shipping this hard? Nothing it stoping you (Besides anthropic) And your mindset Let’s get it Y…
中文: RT @outsource_:伙计们,现在是凌晨3点 你这货运输很辛苦吗? 没有什么能阻止你 (除了人类) 以及你的心态 我们来拿吧 是......
Guys it’s 3 AM Are you shipping this hard? Nothing it stoping you (Besides anthropic) And your mindset Let’s get it YOU CAN JUST BUILD SHIT 🚀 https://twitter.com/outsource_/status/2042866086794670541/photo/1
中文: 伙计们,现在是凌晨3点 你这货运输很辛苦吗? 没有什么能阻止你 (除了人类) 以及你的心态 我们来拿吧 你可以就是建造 SHIT 🚀
🤯 HOW TO GET 4M + IMPRESSIONS IN 2 WEEKS Guys the X algo is opensource Point your agent to the code Ask to what you need to do to go VIRAL I was following this outline without even knowing it. Try : Posting 50 x day Reply 250 x day Drive traffic 📈📈📈 https://twitter.com/outsource_/status/2042864367050383380/photo/1
中文: 如何在2周内获得4M+IMESS X 的家伙是开源的 将你的代理人指向代码 询问你需要做什么才能去病毒 我当时正在遵循这个提纲,甚至都不知道。 试试: 发布50 次日 回复 250 天 驱动流量 📈📈[EE]
It’s Friday night Who’s building?? Who’s drinking 😂
中文: 星期五晚上 谁在建? 谁在喝酒😂
Bullish on Nvidia Need RTX 6000 or H100 👋🏼
中文: 对英伟达的看点 需要RTX 6000或H100 👋🏼
NEW to the Agentic / open source community? I will setup HERMES Agent or OPENClaw for YOU 🫵🏻 For FREE No strings Just wanna help out anyone in need DM ME / Comment 👇🏻👇🏻
中文: 向Agentic/开源社区全新? 我将为您设置HERMES代理或OPENClaw 🫵🏻 免费 无弦 只是想帮助任何需要帮助的人 DM ME / 评论 👇🏻👇🏻
RT @Web3_Matters: Openclaw vs Hermes: Owning your AI Agent Stack https://x.com/i/broadcasts/1dKrPEmbZgyJX
中文: RT @Web3_Matters:Openclaw 与爱马仕:拥有你的AI代理栈
RT @outsource_: LIVE ON X with @cobi_bean @0xQuantic 🧠 https://x.com/i/broadcasts/1YxNrZNzpLLxw
中文: RT @outsource_:与 @cobi_bean @0xQuantic 一起在线直播 🧠
What are you shipping with your OpenClaw // HermesAgent?
中文: 使用您的OpenClaw // HermesAgent发货吗?
Guys the model that launched yesterday is #3 TRENDING on @huggingface Right under NVIDIA and UNSLOTH Gemma 🤯 Check it out 👇🏻
中文: 昨天推出的模特是#3 TRENDING @huggingface 在 NVIDIA 和 UNSLOTH Gemma 的 🤯 下 查看它👇🏻
Great conversation on AI Agents OpenClaw / HermesAgent @ClawSuite & HermesWorkspace If you're on the fence of setting up agents Watch this 👇🏻
中文: 关于人工智能代理的精彩对话 OpenClaw / HermesAgent @ClawSuite & HermesWorkspace 如果你在设立代理人的问题上 观看此视频 👇🏻
HermesAgent inspired MiniMax guys
中文: HermesAgent 激发了 MiniMax 的灵感
LIVE ON X with @cobi_bean @0xQuantic 🧠 https://x.com/i/broadcasts/1YxNrZNzpLLxw
中文: 与 @cobi_bean @0xQuantic 一起在线 🧠
RT @outsource_: 🤯NEW JACKRONG GEMOPUS-4 26-A4B DROPPED The Model Magic 🪄 🧠 Google’s Gemma 4 26B MoE base 4B Active Params 🤖Full Claude O…
中文: RT @outsource_:🤯 新杰克龙 GEMOPUS-4 26-A4B 已停用 模型魔法🪄 🧠 谷歌的 Gemma 4 26B MoE 基础 4B 主动护发 🤖 满满的克劳德·奥......
RT @outsource_: NO WAY https://hermes-workspace.com/ HIT 1,000+⭐️ ITS OFFICIAL 🚀 IN UNDER 1 MONTH We shipped the best UI for HermesAgent We jus…
中文: RT @outsource_: 不, 1000+⭐️ 官方🚀 1个月以下 我们为 HermesAgent 寄送了最佳用户界面 我们只是......
If you are new here say hi for a followback 👋🏻 https://twitter.com/outsource_/status/2042625925242630232/photo/1
中文: 如果你是新来的,请在此处回复👋🏻
No way it crossed 1000 ⭐️ in less than a month..
中文: 不到一个月,它就已超过1000。
No way it crossed 1000 ⭐️ in less than a month.. Building In Public goes HARD
中文: 不到一个月,它就已超过1000。 公共建筑会很难
NO WAY https://hermes-workspace.com/ HIT 1,000+⭐️ ITS OFFICIAL 🚀 IN UNDER 1 MONTH We shipped the best UI for HermesAgent We just released V1 last night! More releases incoming Reply with requested features/ bugs etc 👇🏻 https://twitter.com/outsource_/status/2042624655744290959/photo/1
中文: 不, 以1000+⭐️ 官方🚀 1个月以下 我们为 HermesAgent 寄送了最佳用户界面 昨晚我们刚刚发布了V1! 更多发布内容 使用请求的功能/错误等回复 👇🏻
Follows you goes hard 🦥 https://twitter.com/outsource_/status/2042621263470866559/photo/1
中文: 关注你难点 🦥
Running Agents Through NEW @clawsuite Operations Configure your agents Setup their tasks/ workflows Review their outputs LIVE Help with trend discoveries / trading etc! Polishing it a bit more and will release 🚀 https://twitter.com/outsource_/status/2042618403026817081/photo/1
中文: 通过 @clawsuite 运营来运营代理 配置您的代理 设置他们的任务/工作流程 实时查看他们的输出结果 帮助实现趋势发现/交易等! 再抛光一下,然后发布 🚀
🤯NEW JACKRONG GEMOPUS-4 26-A4B DROPPED The Model Magic 🪄 🧠 Google’s Gemma 4 26B MoE base 4B Active Params 🤖Full Claude Opus-style reasoning distillation Beast Mode Performance🚨 🔥 75 tokens/sec at Q6_K ✅ Just 22.7 GB VRAM 🚨 Full 131k context window Pair this with… https://twitter.com/outsource_/status/2042616365027061797/photo/1
中文: 新杰伦·格莫普斯-4 26-A4B 已下车 模型魔法🪄 🧠 谷歌的 Gemma 4 26B MoE 基础 4B 主动护发液 🤖 完全克劳德·奥普斯式推理提炼 野兽模式性能🚨 🔥 在 Q6_K 处 75 个代币 ✅ 仅 22.7 GB VRAM 🚨 完整的131k上下文窗口 将此与...... 配对
🚨 LOCAL AI JUST HIT THE “OH SH*T” MOMENT This morning I went down a rabbit hole on Gemma 4. ❌ I expected another “small open model, nice benchmarks, still unusable” story ✅ Instead, people are reporting ~40 tok/sec on iPhone, and some HF builds are claiming 45 to 60… https://twitter.com/outsource_/status/2042615891754442824/photo/1
中文: 🚨 本地人工智能刚刚达到“OH SH*T”的时光 今天早上,我在Gemma 4上走下了一个兔子洞。 ❌ 我期待另一个“小幅开放模型、不错的基准测试,仍然无法使用”的故事 ✅ 相反,人们在 iPhone 上报告的是 ~40 次,而一些 HF 版本声称 45 到 60 个......
OPEN MODELS JUST PUNCHED A HOLE IN THE “FRONTIER MOAT” NARRATIVE 🚨 Anthropic dropped Mythos and the story was basically: ❌ only frontier models can do serious AI cybersecurity ❌ the moat is the model Then AISLE tested the same showcase vulns on small, cheap, open-weight… https://twitter.com/outsource_/status/2042613756044157441/photo/1
中文: 开放式模型刚刚在“前护城河”的讲道上拉开了一个洞🚨 人类学掉了神话,故事基本上就是: 只有前沿模型才能进行严重的人工智能网络安全 ❌ 护城河是模型 然后,AISLE 在小巧、廉价、开放重量上测试了相同的展示型钵......
🤯OPUS 4.5 VS OPUS 4.6 PROMPT TEST Users are reporting that when prompted the same thing Opus 4.5 Extended outperforms Opus 4.6 Extended What do you think?? Did Antrhopic NERF Opus 4.6? Is it because of Mythos? View the Results 👇🏻 https://twitter.com/outsource_/status/2042613275934785814/video/1
中文: 🤯 OPUS 4.5 VS OPUS 4.6 PROMPT 测试 用户在提示相同时会报告 Opus 4.5 Extended 表现优于 Opus 4.6 扩展 你觉得怎么样? 无菌的NERF Opus 4.6了吗? 是因为神话吗? 查看结果 👇🏻
🤯OPUS 4.5 VS OPUS 4.6 PROMPT COMAPRSIONS Users are reporting that when prompted the same thing Opus 4.5 Extended outperforms Opus 4.6 Extended What do you think?? Did Antrhopic NERF Opus 4.6? Is it because of Mythos? View the Results 👇🏻 https://twitter.com/outsource_/status/2042613198243721349/video/1
中文: 🤯 OPUS 4.5 对 OPUS 4.6 提示 COMPPRSIONS 用户在提示相同时会报告 Opus 4.5 Extended 表现优于 Opus 4.6 扩展 你觉得怎么样? 无菌的NERF Opus 4.6了吗? 是因为神话吗? 查看结果 👇🏻
Guys now’s your time to SHIP
中文: 伙计们,现在是你去发货的时候了
RT @outsource_: https://hermes-workspace.com/ CROSSED 900 ⭐️ New Release Tonight! Adding a bunch of new features -Profiles -MCP -Context Bar Up…
中文: RT @outsource_: 已突破 900 ⭐️ 今晚新发布! 添加大量新功能 - 配置文件 -MCP - 上下文栏......
RT @MgkMshrmBrkfst: Ladychans and Gentlechads, my magnum opus, Monero --SuperPay-> for Mac is NOW AVAILABLE for download! Hit the fuckin'…
中文: RT @MgkMshrmBrkfst:Ladychans 和 Gentlechads,我的大作 Monero - SuperPay->适用于Mac,现已可下载! 打他妈的......
RT @outsource_: Hermes Workspace just hit v1.0 🚀 (Yes I shipped this from bed) Your AI agent's command center Chat, files, terminal, mem…
中文: RT @outsource_:爱马仕工作区刚刚达到 v1.0 🚀 (是的,我从床上寄了这个) 你的AI代理指挥中心 聊天、文件、终端、我......
Hermes Workspace just hit v1.0 🚀 (Yes I shipped this from bed) Your AI agent's command center Chat, files, terminal, memory, skills, and knowledge browser in one interface. Works with any OpenAI-compatible backend. Ollama, LM Studio, vLLM, Claude, GPT What's new in 1.0:… https://twitter.com/outsource_/status/2042483192968409172/photo/1
中文: 爱马仕工作区刚刚达到了 v1.0 🚀 (是的,我从床上寄来的) 你的AI代理指挥中心 在一个界面中进行聊天、文件、终端、内存、技能和知识浏览器。 适用于任何兼容 OpenAI 的后端。 奥拉马,LM工作室,vLLM,克劳德,GPT 1.0 中有什么新内容:
Hot take Opus 4.6 has been throttled Mythos is running on the same servers as Opus During testing they had to swap compute around Mythos is a HUGE model Only makes sense why we see usage issues & performance dip in Opus 🥱
中文: 热门 OPUS 4.6 已受到限制 Mythos 运行的服务器与 Opus 相同 测试期间,他们不得不将计算进行交换 神话是一个巨大的模型 只有在Opus中看到使用问题和使用问题才有意义🥱
How is this self promotion admin? 😂 Helping people use X to get views? Come on now do better https://twitter.com/outsource_/status/2042474476445511684/photo/1
中文: 这个自助式管理员怎么样?😂 帮助人们使用X来获取视图? 快来,请做得更好
Free sauce if you build in public
中文: 公共场合制作的免费酱料
Free sauce
中文: 免费酱料
✅ How to Go Viral on X in 2026 (Algo Cheat Code) X’s recs algorithm is fully open-source (Grok-powered). It rewards posts that spark REAL conversation + keep people in the app. The winning formula: ⭐️Post 3-5x/day (consistency > volume) 📈Peak times: Tue–Thu, 12–6pm… https://twitter.com/outsource_/status/2042472051412242725/photo/1
中文: ✅ 2026年如何在X上进行病毒式传播(算法作弊代码) X的ress算法完全开源(由Grok驱动)。它奖励那些激发真实对话的帖子,以及让人们留在应用程序中的内容。 获胜公式: ⭐️ 发布 每天3-5次(一致性;数量) 📈 高峰时间:周二–周二至下午12点至6点...
🤯RUN Codex... but 100% LOCAL & completely FREE No subscription No API keys No data leaving your PC Install the official OpenAl Codex CLI locally and run it with Ollama or LM Studio. Dead-simple setup 👇🏻 Ollama route (easiest):✅ 1.npm install -g @openai/codex 2.ollama pull… https://twitter.com/outsource_/status/2042459087539929567/photo/1
中文: 🤯 运行法典......但完全免费 无订阅 无API密钥 没有数据离开你的电脑 在本地安装官方的 OpenAl Codex CLI,并使用 Ollama 或 LM Studio 进行运行。 简单设置 👇🏻 奥拉马路线(最轻松):✅ 1.npm 安装 -g @openai/codex 2. ollama 拉动......
🤯Claude Code… but 100% LOCAL & completely FREE forever No API keys No subscriptions No data leaving your PC Ollama route (easiest):✅ 1.curl -fsSL https://claude.ai/install.sh | bash 2.ollama pull qwen2.5-coder: 32b (or devstral / gemma4) 3.Just run: ollama launch claude… https://twitter.com/outsource_/status/2042458027534450841/photo/1
中文: 🤯 可乐代码......但100% 可自由;完全永久免费 无API密钥 无订阅 没有数据离开你的电脑 奥拉马路线(最轻松):✅ 1. 库尔 -fsSL | bash 2. ollama 拉力 qwen2.5 编码器:32b(或 devstral / gemma4) 3。只需运行:奥拉玛发布会......
🤯Claude Code… but 100% LOCAL & completely FREE forever No API keys No subscriptions No data leaving your PC Ollama route (easiest):✅ 1.curl -fsSL https://claude.ai/install.sh | bash 2.ollama pull qwen2.5-coder: 32b (or devstral / gemma4) 3.Just run: ollama launch claude… https://twitter.com/outsource_/status/2042457758969057701/photo/1
中文: 🤯 可乐代码......但100% 可自由;完全永久免费 无API密钥 无订阅 没有数据离开你的电脑 奥拉马路线(最轻松):✅ 1. 库尔 -fsSL | bash 2. ollama 拉力 qwen2.5 编码器:32b(或 devstral / gemma4) 3。刚刚跑:奥拉马发布会......
🤯Claude Code… but 100% LOCAL & completely FREE forever No API keys No subscriptions No data leaving your PC Run the real Claude Code VS Code extension with Ollama or LM Studio in 2 minutes. Dead-simple setup 👇🏻 ✅Ollama route (easiest): 1. Install Ollama →… https://twitter.com/outsource_/status/2042456927880237519/photo/1
中文: 🤯 可发代码......但100% 可自由;完全永久免费 无API密钥 无订阅 没有数据离开你的电脑 使用 Ollama 或 LM Studio 运行真正的 Claude Code 与 S Code 扩展程序,只需 2 分钟。 简单设置 👇🏻 ✅ 奥拉马路线(最简单): 1。安装Ollama →...
We sent her a little to hard boys Anyone experiencing Usage issues? Max 20x ran out of usage in like 1 hr lol GGS https://twitter.com/outsource_/status/2042445705277567022/photo/1
中文: 我们给她送了一小会生的男孩 有人遇到使用问题吗? 最高20倍的用量与1小时的GGS一样用完了
RT @outsource_: How many people started using HermesAgent or OpenClaw from my posts? Or other posts on X? Where did you find out about it…
中文: RT @outsource_:有多少人开始使用我发帖的 HermesAgent 或 OpenClaw? 还是X上的其他帖子? 你从哪儿知道的......
OpenAI > Claude LOL
中文: OpenAI > 克劳德 哈哈
RT @outsource_: 🤯NEW https://hermes-workspace.com/ MOBILE COOKS! The BEST command center for your Hermes Agent! One unified workspace. Zero t…
中文: RT @outsource_:🤯 NEW 移动 COOK! 最适合您爱马仕代理的指挥中心! 统一的工作空间。零......
RT @outsource_: BREAKING: @ClawSuite COOKED OPENCLAW FOR MOBILE 🔥 This is how OpenClaw Mobile should be 🤯 The BEST interface to run Open…
中文: RT @outsource_: 爆料:@ClawSuite 为手机 🔥 烹饪 OPENCLAW 这就是 OpenClaw Mobile 应该如何实现的🤯 运行Open...的最佳界面
🤯NEW https://hermes-workspace.com/ MOBILE COOKS! The BEST command center for your Hermes Agent! One unified workspace. Zero tab/terminal chaos. Full power at your fingertips: 🤖 Chat + live tool execution 🧠 Memory browser ⭐️Skills catalog (100+) 🚀 Built-in Terminal 🌐File… https://twitter.com/outsource_/status/2042411498081866175/video/1
中文: 🤯 来源: 移动式 COOK! 最适合您爱马仕代理的指挥中心! 统一的工作空间。零标签页/终端混乱。 指尖全功率: 🤖 聊天 + 实时工具执行 🧠 内存浏览器 ⭐️技能目录(100+) 🚀 内置终端 🌐 资料图......
Check out HermesWorkspace!
中文: 查看HermesWorkspace!
My biggest project is still growing!
中文: 我最大的项目还在不断发展!
If you're running OpenClaw you NEED this
中文: 如果你正在运行 OpenClaw,你需要这个
If you're running OpenClaw you NEED this
中文: 如果你正在运行 OpenClaw,你需要这个
BREAKING: @ClawSuite COOKED OPENCLAW FOR MOBILE 🔥 This is how OpenClaw Mobile should be 🤯 The BEST interface to run OpenClaw on your phone! 🚀Full ClawSuite access in your pocket: • 📈Dashboard ⭐️Skills 🤖Chat 🧠Operations & Everything else! ALL your agent sessions on… https://twitter.com/outsource_/status/2042409763598749910/video/1
中文: 爆料:@ClawSuite 为手机 🔥 烹饪 OPENCLAW 这就是 OpenClaw Mobile 应该如何实现的🤯 在手机上运行OpenClaw的最佳界面! 🚀 口袋里的全包式服务 📈 仪表板 ⭐️ 技能 🤖 聊天 🧠 操作 安培;其他一切! 您的所有代理会议都在以下网址进行...
https://hermes-workspace.com/ CROSSED 900 ⭐️ New Release Tonight! Adding a bunch of new features -Profiles -MCP -Context Bar Updates -Knowledge Graph etc! Check it out 👇🏻 https://github.com/outsourc-e/hermes-workspace https://twitter.com/outsource_/status/2042409177511838170/photo/1
中文: 已 900 查询 ⭐️ 今晚新发布! 添加大量新功能 - 配置文件 -MCP - 上下文条更新 知识图谱 等等! 快来看看👇🏻
How tf did they ship this without Claude?? https://twitter.com/outsource_/status/2042399266484720014/photo/1
中文: 没有克劳德,他们是怎么寄来的?
RT @outsource_: 🚨REQUESTED MLX TUNE DGEMMA 4-31B LANDED 🦥@UnslothAI native 4-bit MLX for Apple Silicon 🦥 🔥Blazing fast inference on al…
中文: RT @outsource_:🚨 请求MLX TUNE DGEMMA 4-31B LANDED 🦥@UnslothAI 原生 4 位 MLX 用于 Apple Silicon 🦥 🔥 在 al. 上快速推断
RT @outsource_: 🚀 NEW GEMMA 4 31B TURBO DROPPED Runs on a SINGLE RTX 5090: ⚡️18.5 GB VRAM only (68% smaller) 🧠51 tok/s single decode 💻1,…
中文: RT @outsource_:🚀 全新 GEMMA 4 31B TURBO 已停车 单次RTX 5090上运行: ⚡️18.5 GB 仅 VRAM(缩小 68%) 🧠51 单次解码 💻1,......
This is very interesting honesty. But the main concern is still there! These benchmarks dont prove anything. On paper it shows data But whats the actual use cases? Perfomance? Tool Calling? Real benchmarks are still needed. Will fix this
中文: 这非常有趣。 但主要担忧仍然存在! 这些基准指标无法证明任何情况。 纸面上显示数据 但实际使用案例有哪些? 香水?工具调用? 仍然需要真正的基准。 会解决这个问题的
🚨REQUESTED MLX TUNE DGEMMA 4-31B LANDED 🦥@UnslothAI native 4-bit MLX for Apple Silicon 🦥 🔥Blazing fast inference on all M-series Macs 🤖Super efficient (~20GB RAM only) 🤯Strong multimodal + vision performance 🌐Full 256K context + native function calling 🔥Crushes… https://twitter.com/outsource_/status/2042383535659466934/photo/1
中文: 🚨 请求MLX TUNE DGEMMA 4-31B 上路 🦥@UnslothAI 原生 4 位 MLX 用于 Apple Silicon 🦥 🔥 在所有M系列 Mac上快速推断 🤖 超高效(仅 ~ 20GB 内存) 🤯 强大的多模态+视觉性能 🌐 完整的 256K 上下文 + 原生函数调用 🔥 Crushes...
How many people started using HermesAgent or OpenClaw from my posts? Or other posts on X? Where did you find out about it?? Comment down 👇🏻👇🏻
中文: 有多少人开始从我的帖子中使用 HermesAgent 或 OpenClaw? 还是X上的其他帖子? 你在哪里发现的? 评论一下👇🏻👇🏻
RT @outsource_: 🚨BREAKING: NEW HERMES AGENT QWEN MODEL @CardilloSamuel drops Carnice-MoE-35B-A3B Inspired by @kaiostephens Carnice Mode…
中文: RT @outsource_:🚨 突发事件:全新赫姆斯 AGENT QWENE 模特 @CardilloSamuel 放弃 Carnice-MoE-35B-A3B 灵感来自@kaiostephens Carnice模式......
OPENSOURCE FTW @LilaRest COOKED🔥
中文: OPENSOURCE FTW @LilaRest 已 COOKED🔥
Check out the release 👇🏻👇🏻
中文: 查看发布 👇🏻👇
@LilaRest COOKED with this 🔥🔥 https://x.com/lilarest/status/2042320271005069618?s=46&t=CwqMMkPqtrbYKjkY3iPXbw
中文: @LilaRest 已使用此 🔥🔥 进行 COOK
🚨BREAKING: NEW HERMES AGENT QWEN MODEL @CardilloSamuel drops Carnice-MoE-35B-A3B Inspired by @kaiostephens Carnice Model What it is: 🚨35B MoE model (~3B active params only) 🔥Qwen 3.5 base model optimized for agentic work ⚡️Built for god-tier agentic performance What it… https://twitter.com/outsource_/status/2042351974549766530/photo/1
中文: 突发事件:新赫默斯经纪人量化宽松模式 @CardilloSamuel 放弃 Carnice-MoE-35B-A3B 灵感来自@kaiostephens Carnice模型 是什么: 🚨35B MoE 模型(仅 ~3B 活性参数) 🔥 针对特效工作优化了 3.5 基础模型 ⚡️ 专为神性级特效而生 什么......
RT @outsource_: 🚀 NEW GEMMA 4 31B TURBO DROPPED Runs on a SINGLE RTX 5090: ⚡️18.5 GB VRAM only (68% smaller) 🧠51 tok/s single decode 💻1,…
中文: RT @outsource_:🚀 全新 GEMMA 4 31B TURBO 已停车 单次RTX 5090上运行: ⚡️18.5 GB 仅 VRAM(缩小 68%) 🧠51 单次解码 💻1,......
43,000+ ⭐️ INSANE!!!!!
Really good AI news 👇🏻👇🏻
中文: 真正好的人工智能新闻👇🏻👇
The second I realized nobody really knows anything about the AI space I realized I could cook. Nobody knows everything It’s up to you to research -> test -> publish Start using agents/ harnessss Start breaking shit Start developing things you wish existed Go all in. Now’s the… https://twitter.com/outsource_/status/2042326230431908087/photo/1
中文: 第二次我意识到,我真正了解的人工智能空间是什么,我意识到自己能做饭。 没有人知道一切 由你来研究 -> 测试 -> 发表 开始使用代理/线束 开始搞砸 开始开发你希望存在的东西 全场去。 现在是......
nice I just reposted
中文: 太好了,我刚刚转发了
🚀 NEW GEMMA 4 31B TURBO DROPPED Runs on a SINGLE RTX 5090: ⚡️18.5 GB VRAM only (68% smaller) 🧠51 tok/s single decode 💻1,244 tok/s batched 🤖15,359 tok/s prefill ← yes, fifteen thousand 🚨2.5× faster than base model with basically zero quality loss. It hits Sonnet-4.5… https://twitter.com/outsource_/status/2042323470428537121/photo/1
中文: 🚀 新款 GEMMA 4 31B TURBO 已下降 单次RTX 5090上运行: ⚡️18.5 GB 仅 VRAM(缩小 68%) 🧠51 单次解码 💻1,244 次批次 [EE4/15,359 次预填 ← 是,一万五千 🚨2.5× 比基础模型快,质量损失基本为零。 点击Sonnet-4.5......
I do this already with OpenClaw / Hermes. If you aren’t running Agentic loops To validate work / review output You’re NGMI
中文: 我已经使用 OpenClaw / Hermes 进行此事了。 如果你没有运行 Agentic 循环 验证工作/评审输出 你是NGMI
As Anthropic blunders PR OpenAI takes the lead Better community / support More user focused Ships things users request At this rate if they keep this up And improve their models OPENAI > CLAUDE But at the moment Opus 4.6 is still better than GPT 5.4
中文: 因为人类在公关上失误 OpenAI 领先 更好的社区/支持 更加注重用户 用户需要的船舶 以这种速度,如果他们继续这样做 并改进他们的模型 奥普奈 > 克劳德 但目前Opus 4.6仍优于GPT 5.4
THANK YOU
中文: 谢谢
DMS Open 👋🏼
中文: DMS 开放 👋🏼
RT @outsource_: 🚨NEW Features dropping for https://hermes-workspace.com/ 🧠 Built In Knowledge View + Knowledge Graph ⚡ MCP Server Config 👥 Profi…
中文: RT @outsource_:🚨 新功能下降 🧠 内置于知识视图+知识图谱中 ⚡ MCP 服务器配置 👥 专业......
Update boys they got us Claude OAUTH in hermes is cooked https://twitter.com/outsource_/status/2042088380188528878/photo/1
中文: 更新他们给我们的 克劳德·非统组织在爱马仕中的制作内容为
If you're on the fence about setting up OpenClaw // HermesAgent Or are running it and have questions Comment + DM ME for help ⭐️
中文: 如果你在设置 OpenClaw // HermesAgent 时持谨慎态度 或者正在运行它并有疑问 评论 + DM ME 寻求帮助 ⭐️
Claude could never 😂
中文: 克劳德永远无法[EE]
🚨 MICROSOFT NEW BITNET.CPP ALLOWS YOU TO RUN 100B MODELS LOCALLY 🔥Official inference framework for 1-bit LLMs. It lets you run massive 100 billion parameter models on a regular CPU Human reading speed — 5–7 tokens/sec Zero GPU needed.• Up to 6.17× faster inference on… https://twitter.com/outsource_/status/2042067441958691091/video/1
中文: 🚨 MICROSOFT NEW BITNET.CPP 可让您在本地运行 100B 模型 🔥 用于1位LLM的官方推理框架。 它允许你在普通CPU上运行高达1000亿的参数模型 人类阅读速度——5–7 个代币 需要零 GPU。• 最高可快速推断 . . .
New features dropping 👀
中文: 新功能下降👀
🚨NEW Features dropping for https://hermes-workspace.com/ 🧠 Built In Knowledge View + Knowledge Graph ⚡ MCP Server Config 👥 Profiles - Manage all Hermes Profiles 🔌 Vanilla Hermes compatibility & MORE! DROPPING SOON 🤯 https://twitter.com/outsource_/status/2042055668253721073/photo/1
中文: 🚨 新功能下降 🧠 内置于知识视图+知识图谱中 ⚡ MCP 服务器配置 👥 个人资料 - 管理所有爱马仕个人资料 🔌 香草香草相容性 加以;更多!即时下车🤯
What theme are you using in your HermesWorkspace? https://twitter.com/outsource_/status/2042051728527257943/photo/1
中文: 你在 HermesWorkspace 中使用什么主题?
Its like Christmas morning 😂 Guys check you Claude Usage Usage JUST RESET LETS GOOOOO https://twitter.com/outsource_/status/2042046998249918615/photo/1
中文: 就像圣诞节的早晨😂 男生们会检查你的克劳德使用情况 使用方法 只是 RESET LETS GOOOO
Still bullish on CLAUDE 😂
中文: 对克劳德😂仍然持乐观态度
I keep going back to Opus
中文: 我不断回到《行动会
🚨 HERMES AGENT v0.8.0 JUST DROPPED — “THE INTELLIGENCE RELEASE” IS HERE 🔥 The most dangerous open-source AI agent just leveled up HARD: • Background tasks now auto-notify you when done (notify_on_complete) 📣 • FREE MiMo v2 Pro on the Nous Portal • Live model switching…
中文: 🚨 HERMES AGENT v0.8.0 刚被放弃——“情报释放”到此处 🔥 最危险的开源人工智能代理刚刚升级了: • 后台任务在完成时会自动通知您(notify_on_complete)📣 • 在诺斯门户上免费使用 MiMo v2 Pro • 实时模型切换...
🚨 Anthropic just dropped Claude Managed Agents today… Open-source devs answered in hours. Meet Multica the fully open (Apache 2.0) version that turns Claude Code + OpenAI Codex agents into your actual autonomous teammates. ✅ Assign GitHub issues ✅ They plan, code &…
中文: 🚨 人类学今天刚刚放弃了克劳德管理代理公司...... 开源开发人员在数小时内得到了回复。 与Multica一起实现完全开放(Apache 2.0)版本,将Claude Code+ OpenAI Codex代理转变为真正的自主型团队成员。 ✅ 分配 GitHub 问题 ✅ 他们计划,代码与编程;......
RT @Scobleizer: https://twitter.com/i/spaces/1DGleEpaXLrJL
https://hermes-workspace.com/ will be merged to main very soon guys! Making some changes to the way we connect to HermesAgent Will no longer need my agent fork very soon.. Letting my agent Aurora Cook this 🚀 https://twitter.com/outsource_/status/2042015535823479066/photo/1
中文: 将很快合并为主人! 对我们与 HermesAgent 的连接方式进行一些更改 很快就不再需要我的经纪人了。 让我的经纪人Aurora Cook 这个 🚀
Bullish on Hermes
中文: 对爱马仕的欺凌
What’s the best open model for your use case, and why? 👇🏻👇🏻 Local or Cloud?
中文: 最适合您使用的开放式型号是什么?为什么?👇🏻👇🏻 本地还是云?
Shopify CEO bullish on Hermes 😬
中文: Shopify首席执行官对爱马仕持乐观态度 😬
Claude on a generational run They are shipping faster than any company Why?? Cause they literally engineered the best model with Mythos Their engineers understands the game They are using it to ship faster than anyone.
中文: 克劳德在一代人的奔跑中 它们的发货速度比任何公司都快 为什么? 因为他们确实用Mythos设计了最佳模型 他们的工程师理解这款游戏 他们正用它来比任何人都更快地发货。
We are in a bubble on X
中文: 我们正处于X的一个泡沫中
You got a model you get a model https://twitter.com/outsource_/status/2041927656678039619/photo/1
中文: 你有一个模型,你得到一个模型
What a time to be a hacker 💻
中文: 成为一名黑客的时机💻
Checkout HermesAgent Skins guy 🤯
中文: 查看 HermesAgent 皮肤 家伙 🤯
Guys run this if you have 366263GB VRAM
中文: 如果你有366263GB的VRAM,这些人就会运行这个
All these new OpenSource models release What benchmarks are useful? What tk/s can you run?? What can my hardware run? We need answers for these questions
中文: 所有这些新的开源模型都会发布 哪些基准是有用的? 你能跑什么? 我的硬件能运行什么? 我们需要这些问题的答案
HERMES-AGENT WORKS WITH CLAUDE OAUTH 🔥 While everyone else is hacking together proxies and workarounds to get Claude working locally... @NousResearch quietly built it right into Hermes. No proxy needed. No patches. No hacks. You just: → Run claude auth login → Point… https://twitter.com/outsource_/status/2041005554395390384/photo/1
中文: 赫默斯-阿肯特与克劳德·奥瓦蒂合工 🔥 其他人正在一起入侵代理和变通方法,以让克劳德在本地工作...... @NousResearch 悄然将其构建成 Hermes。 无需代理。没有补丁。没有黑客。 你只是: → 运行 claude 登录 → 要点......