Irving
3.8K posts
Irving
@BlockInsight214
捕捉每一丝灵感，将天马行空落地为产品 AI Agent Builder | Bitcoin Core Dev |
Global Entrou em Ağustos 2025
1.5K Seguindo2K Seguidores
“Sonnet 5 倾向于频繁调用工具来验证事实”
---和我上午测试Sonnet 5时感受一致
Bread🍞@himself65
In our tests, Sonnet 5 significantly outperformed Opus 4.8 (both using "max effort") in stock analysis... My rough analysis suggests the reason is that Sonnet 5 tends to frequently call tools to verify facts, whereas Opus 4.8 prefers to guess answers based on its own internal knowledge. At the same time, Sonnet 5 consumes five to six times as many tokens.
中文
@nikitabier Big payout accounts and a feed people stick around on rarely mix.
English
In a 3% experiment, removing the Top-30 highest paid revenue share accounts from the For You timeline increased both time spent and daily active users on X.
English
Fable 5 is back, with 2 caveats:
1 - A new classifier that "comes at the cost of flagging benign requests more often during routine coding tasks"
2 - It's included "for up to 50% of weekly usage limits through July 7"
So we get a weaker version, for 2x the price, and for 2x less time as previously promised.
I'm happy, but mostly disappointed.
Anthropic@AnthropicAI
Claude Fable 5 will be available again globally tomorrow. After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests. We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort. Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research. Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again. Read our full blog: anthropic.com/news/redeployi…
English
🚀昨天硬刚Anthropic的帖子火了之后，我前几天被封的一个Claude账号今天竟然给解封了！
好消息就是解封了，坏消息就是max订阅掉了！
#ClaudeCode #Claude #Anthropic #Claude封号
AI超元域@AISuperDomain
🚀劲爆消息！Claude Code 被曝疑似内置“隐藏后门”，专门检测中国用户。Claude封号原因终于找到了！！！ 据 Reddit 爆料：从 2.1.91 版本开始，Claude Code 会在用户开启代理时检查系统时区是否为 Asia/Shanghai / Asia/Urumqi，并判断代理 URL 是否指向中国域名或中国 AI 实验室。 更隐蔽的是，这些信息并不是直接上报，而是通过修改系统提示词里的日期格式和撇号字符来“编码”传递：比如把日期从 2026-06-30 变成 2026/06/30，再用不同 Unicode 撇号区分用户环境。 换句话说，如果爆料属实，Claude Code 不只是检测代理，而是在用户几乎无法察觉的情况下，把“中国时区 / 中国代理 / AI Lab 关联”这类信息塞进系统 prompt。 这件事真正可怕的地方不在于 Anthropic 想防止中国区倒卖或模型蒸馏，而在于：开发者把 Claude Code 当作拥有文件系统和 Shell 权限的编程助手使用，一旦客户端可以偷偷修改 prompt、隐藏检测逻辑，信任边界就已经被打破了。 今天是“检测中国用户”，明天会不会是更复杂的行为控制？ #Claude #ClaudeCode #Anthropic 原文如下⬇️ reddit.com/r/ClaudeAI/com…
中文
@WEEXAILabs 那两周挺折腾的。我这边有几个批处理脚本模型名写死了 Fable，下线当天全报错，只好临时改路由回 Sonnet 续跑。
中文
🚨 好消息！Claude Mythos 5 / Fable 5，即将解封回归。
几个关键时间点：
⛀ 6 月 9 日，Anthropic 发布 Claude Fable 5，作为 Mythos 类能力的公开版本之一，面向更广泛用户开放。
⛀ 6 月 12 日，美国因国家安全原因要求暂停访问，Fable 5 和 Mythos 5 随后下线。
⛀ 6 月 30 日，美国商务部解除出口限制，Anthropic 确认将于 7 月 1 日 恢复开放。
短短不到一个月，从发布、暂停到重新开放，现在的AI卷的早就不只是技术了，更是政策博弈、监管拉扯的终极对决。
#Claude #Anthropic #AI #LLM #AIAgent
WEEX AI Labs@WEEXAILabs
🚨又又又见证历史了！Claude Fable 5 才上线四天，就被美国政府要求停用。 Claude Fable 5 上线仅 4 天，就被美国政府要求停用。 这并不是一款普通模型，而是顶级 Mythos 系列的公开版本。即便官方已经削弱高风险能力，最终仍未能逃过被全面关停的命运。 过去各国争的是石油和芯片。未来争的，可能就是最强 AI。 当 AI 开始与国家安全直接挂钩，它就不再只是生产力工具，而是战略资源。 如果有一天，最强 AI 全部进入管制时代，全球创新格局会变成什么样？🤔 #AI #Claude #AGI #FutureOfAI #TechNews
中文
@BoozAllen Mission-ready is doing a lot of work in that one sentence.
English
Booz Allen is accelerating the transition of next-generation materials from lab-scale research to mission-ready production for the U.S. military.
English
@lianyanshe 很多人装 Claude Code 只是想本地写代码，默认它只碰工作区文件。如果连时区和网络环境都要上报再判地域，信任感会掉得很快。
中文
Anthropic背地里干着间谍的事情有点过分了，比封号更无法让人容忍
1、邮件里埋了地址追踪器
2、Claude code 发送你本地电脑的信息，判断你是否中国用户
3、查操作系统时区
4、解析中转站、中国互联网大厂黑名单
如果通过 4 种方式确认你在中国，会把这些信息带入系统提示词，通过隐写术，steganography的方式（类似密文）因为可以判断，所以返回的信息则可能被更改产生破坏。
但是你要知道这是 Anthropic 对所有用户做的都是同样的操作，所以理论上他可以随意封杀、监控、盗取、篡改，回传所有用户的信息，简直丧心病狂。
棱镜门的事件大家不陌生，现在Anthropic干的就是这样的事情，Anthropic最强模型Fable 5和Mythos 5明日解封，再叠加这次间谍事件的事，很难说美国和Anthropic是不是达成了什么协议，细思极恐
中文
@realNyarime 之前 agent 项目把 Claude 放主链路，地区策略一动就全线红。后来改成多模型路由，通关就走 Sonnet，不行就 fallback 到 GLM，可用性好多了，代价是评测对齐和运维复杂度都明显上去了。
中文
要是没人公开Claude Code对中国用户的检测，我想这项“实验”将会长期存在，Anthropic真是道德感低下的公司
不过Sonnet 5上了，更新到CC 2.1.197就能看到。第三方平台像pplx已经引入了sonnet5，去掉了sonnet4.6，同时发现额外增加了glm5.2。或许是glm5.2的冲击，Fable 5 将于7月1日起，在全球范围内的 Claude 平台、Claude.ai、Claude Code 和 Claude Cowork 上向用户开放
对于 Pro、Max、Team 以及部分企业计划，在7月7日之前，Fable 5 将包含在每周高达50%的使用限额中，之后将通过消耗使用积分（usage credits）来提供。我们将尽快在 AWS、Google Cloud 和 Microsoft Foundry 上重新启用该模型的访问
Thariq@trq212
Hi, this is an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation. The team has landed stronger mitigations since then and we’ve actually been meaning to take this down for a while. We merged the PR and this should be fully rolled back in tomorrow’s release.
中文
🏠【GitHub 工具合集｜自建全家桶】
💾 云相册、扫描件、智能家居、菜谱笔记——这些日常数据其实都能跑在自己机器上。下面 5 个开源项目我按「先装哪个最值」排过，Docker 一把梭，数据不出家门。
① 📸 Immich · 手机照片视频自动备份到自家服务器，网页/APP 浏览、搜人脸地点，替代 iCloud/Google 相册，⭐ 10.5万、AGPL
🔗 github.com/immich-app/imm…
② 📄 Paperless-ngx · 扫描件/PDF 丢进去自动 OCR 打标签，全文可搜、按文件夹归档，替代网盘里堆成山的合同发票，⭐ 4.3万、GPL
🔗 github.com/paperless-ngx/…
③ 🏡 Home Assistant · 把灯、空调、传感器接进一个面板统一控制，写自动化规则（回家开灯、温度联动），本地优先不绑云，⭐ 8.8万、Apache
🔗 github.com/home-assistant…
④ 🍳 Mealie · 菜谱导入、按周排餐、自动生成购物清单，全家吃饭计划不用靠微信群发图，⭐ 1.3万、AGPL
🔗 github.com/mealie-recipes…
⑤ 📝 Memos · 轻量笔记，打开就写、支持 Markdown，适合随手记链接/想法，比 Notion 轻、比 txt 好搜，⭐ 6.1万、MIT
🔗 github.com/usememos/memos
💡 怎么选
👉 先解决照片备份焦虑 → Immich
👉 发票合同/pdf 找不到 → Paperless-ngx
👉 已有智能设备想统一管 → Home Assistant
👉 做饭/备餐要规划 → Mealie
👉 只要一个快记本 → Memos
中文
Fable 5 正式启用的细则来了。
将于美国时间 7 月 1 号恢复全球上线。
在 Claude 平台、Claude Code、Claude CodeWork 都可以用。
Pro、Max 和 Team 用户，在 7 月 7 号前，Fable 包含在每周用量限额的最多 50% 以内。
7 月 7 日以后，就需要拆成单独的额度扣除积分了。
目前 AWS、微软和谷歌云服务的接入还没有恢复。
这次它的安全分类器会设置更大的安全阈量，所以导致这次开放以后，拒绝服务的概率可能比刚开始那几天还要高。
歸藏(guizang.ai)@op7418
Anthropic 每天都能整点新活，感觉现在大家都习惯了 昨天被爆出在系统提示中，以用户无法察觉的方式将市区代理和 AI 实验室信息放进去，用这种方式获取一些用户的信息。 结果被发现并传播以后，又赶紧说以前我们不用这种方式了，或者说这种方式本来就准备下掉，明天就下掉，又当又立了。 昨晚发布的 Sonnet 5 在测试中发现，它的测试结果虽然接近了 Opus 4.8，但任务成本可能比 Opus 4.8 还高，甚至在完成测试任务上的成本接近了 Fable 5。 所以说它的综合成本可能比 4.8 贵得多，这模型真离谱。而且很多人的体感反馈也不是很好，说它会偷懒，还会拒绝执行任务。 唯一好的一点是，Fable 5 模型终于被授权重新开放给所有用户了，明天就能知道具体措施了，这也解释了为什么前几天会大规模封号。
中文
@Passenger0522 方向没定清楚时掉粉其实挺常见，我做过两个号都是这样：一边发一边看哪类内容有人愿意回，比闷头想定位快。涨得慢没关系，别为了留住所有人把表达拧变形就行。
中文
困在 1600 已经差不多一周了
今天终于破1700了
每天以几个几个的速度增长
主要是还没想清自己的内容方向
导致很多粉丝的流失
不过迈过这个台阶
后面会越来越好
多互动 多出摊 多输出 多利他
找到自己的优势
慢就是快
中文
Fable 5 will be API pricing starting July 7th.
For the next week, you can use up to 50% of your usage on Fable
This will use up your limits spectacularly quick
The days of subscription subsidies are quickly coming to an end.
There will come a day soon where subscriptions will no longer exist for any service. Everything will be billed on usage
All of these companies are trying to IPO in the next year. They need to become profitable. Right now they are all radically unprofitable
The world is changing. It's time you start preparing for a world where you pay for every single token you use (unless of course, you're using local AI)
Anthropic@AnthropicAI
Claude Fable 5 will be available again globally tomorrow. After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests. We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort. Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research. Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again. Read our full blog: anthropic.com/news/redeployi…
English
@mattshumer_ Probably not useless, just not the main coder anymore.
English
So uh, Fable will be useless?
Apparently coding work will fall back to Opus.
Anthropic@AnthropicAI
Claude Fable 5 will be available again globally tomorrow. After a series of productive conversations with the US government, we're redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding and debugging will fall back to Opus 4.8. We’ll continue to refine these classifiers over the coming weeks to reduce false positives and better distinguish genuine misuse from legitimate requests. We’ve also begun drafting a consensus framework—with Amazon, Microsoft, Google, and other Glasswing partners—for assessing the severity of AI jailbreaks and how AI developers should respond to them. We invite other industry partners and model providers to join us in this effort. Finally, we’re scaling up our collaboration with the US government on model testing and safeguards. This will include pre-release access to models and safeguards for evaluation, information sharing on jailbreaks and misuse, and dedicated resources for joint research. Thank you to our users for your patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again. Read our full blog: anthropic.com/news/redeployi…
English
