Chrix @chrixbell

Color Obsession | 自幼好色 | HP-Discovery Joined May 2009

Tweets

1K
Followers

105
Following

873
Likes

145

Chrix @chrixbell

2 weeks ago

@zstmfhy 不能重GPT API credit

0 0 0 101 0

View Details

Arya@羊姐社区🦅 @Arya_web3

3 weeks ago

132 39 753 60K 65

View Details

Chrix @chrixbell

4 weeks ago

@xdd1458 Gemini restored

0 0 1 82 0

View Details

刘小排 @bourneliu66

a month ago

一件非常恐怖的事，悄然发生了： Qwen3.8-27B 在大部分能力上都超过或逼近了Claude Opus 4.6 Max！也就是说，从今天开始，你在自己的电脑上、单机单卡能够部署的小模型，已经拥有了不逊于Claude Opus 4.6(半年前的世界之巅）的智能！无论是写代码、还是驱动OpenClaw ！

157 61 644 166K 400

View Details

Fact @Fact

a month ago

If You Can’t Convince Them, Confuse Them - Harry Truman

7 70 640 71K 64

View Details

南极熊 @DDT34

a month ago

挨饿这事如果干的好就叫减肥；掐人这事如果干的好就叫按摩；发呆这事如果干的好就叫深沉；偷懒这事如果干的好就叫享受；死皮赖脸这事干的好就叫执着。

0 2 2 147 0

View Details

禿道道🐟 @dearemon

2 months ago

果然，出门在外身份都是自己给的

187 108 4K 347K 226

View Details

Web x @Thank1519

2 months ago

不要碰任何贫穷的东西，包括人。贫穷的背后是一整套麻烦系统，认知混乱、边界缺失、情绪不稳定、亲情勒索、道德绑架、喜欢占便宜，习惯性索取。

0 6 39 3K 1

View Details

@minchoi Our 2T model, which is better than our 1.5T in every way, will finish initial training next week. It might be able to exceed Kimi, but with speed and token efficiency close to our 1.5T (aka Grok 4.5).

693 764 12K 1.8M 687

View Details

Claude @claudeai

2 months ago

Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits. Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit. Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity.

5K 6K 53K 29.0M 7K

View Details

Chrix @chrixbell

3 months ago

Goal Achieved

0 0 1 12 0

View Details

SemiAnalysis @SemiAnalysis_

3 months ago

With the introduction of the TPUv8t, their new training focused TPU, Google unveiled a new scale-out network architecture called Virgo. Virgo is able to interconnect up to 134,400 chips with up to 47 Pbps of non-blocking bi-sectional bandwidth. (1/4)🧵

9 49 458 86K 234

View Details

Chrix @chrixbell

4 months ago

@itopinion 改成了Codex github.com/chrixbell/eink… 参考自claude-eink-bridge github.com/BarryBarrywu/c…

0 1 0 39 0

View Details

Chrix @chrixbell

4 months ago

@seclink 代持，合法

0 0 0 65 0

View Details

Chrix @chrixbell

5 months ago

@itopinion 下单支持了

0 0 0 78 0

View Details

Max Lv @m0d8ye

5 months ago

众所周知， OpenRouter 上按token消耗量排名，全球前六大模型均为中国产品，中国模型整体占全球用量近半。但更值得关注的是支撑它的物理基础设施——顺着资金流和算力部署路径追踪，指向的是同一个地理坐标：新柔巴（SIJORI）走廊。触发这轮布局的直接变量是美国《远程访问安全法案》。此前，中国企业可通过第三国云服务商远程租用高端GPU集群训练模型，不涉及实体芯片越境，在既有出口管制框架下属于灰色地带。该法案拟将远程访问受控芯片的行为正式纳入管制，一旦生效，这条轻资产算力获取通道将被关闭。对中国AI企业而言，自建海外算力底座从战略选项变成了生存刚需。选址逻辑上，新柔巴走廊的优势在于三地形成了高度互补的功能分工。新加坡作为亚太最大海缆登陆中心和金融枢纽，承担资金结算与流量调度功能；柔佛工业电价约为新加坡的三分之一，土地储备充裕，仅字节跳动一家的投资承诺已超62亿美元，柔佛整体数字基础设施规划投资逾380亿美元；巴淡岛距新加坡仅20公里，兼具算力溢出承接和印尼本土市场入口的双重价值。这套架构同时覆盖了合规身份、运营成本和市场准入三个关键维度。资本通道的打通同样重要。智谱AI与MiniMax今年1月先后在港交所上市，分别募资逾43亿和48亿港元。智谱招股书显示，其去年上半年研发支出近16亿元人民币，其中逾11亿元为第三方算力采购费用。在融资架构层面，有些中国公司像是万国数据将海外业务剥离为新加坡注册的DayOne，通过引入软银、高瓴及印尼主权基金等国际资本，将持股稀释至约24%并完成出表。DayOne由此获得独立的国际化法律身份，顺利取得35.4亿美元贷款，芯片采购亦不再面临合规障碍。这一”去中国化”重组路径正在被更多企业参考。回到开头的数据。中国模型能以近半份额主导全球token消耗，核心驱动力是定价——通常为美国同类产品的十分之一至二十分之一。这种价差的底层支撑并非仅靠算法效率，更是能源经济学层面的结构性优势：东南亚低成本电力经GPU运算转化为token输出后，单位电力的经济价值可提升约22倍。综上可以看到一条完整的价值链条：出口管制收紧倒逼重资产出海，SIJORI走廊以综合禀赋承接算力落地，资本通过架构重组疏通芯片与融资管道，最终以极致成本优势的token输出占领全球开发者市场。当这套闭环成型，竞争焦点将从模型能力本身转向能源获取与转化效率。

21 44 297 134K 304

View Details

Akshay 🚀 @akshay_pachaar

5 months ago

A raw LLM is just like a CPU without OS. It can compute. But it can't do anything useful on its own. This analogy is the clearest way I've found to understand what an agent harness actually does. Here's the mapping: • 𝗖𝗣𝗨 → 𝗟𝗟𝗠 (model weights). The raw compute engine. Powerful, but useless without infrastructure around it. • 𝗥𝗔𝗠 → 𝗖𝗼𝗻𝘁𝗲𝘅𝘁 𝘄𝗶𝗻𝗱𝗼𝘄. Fast, always available, but limited. When it fills up, you start losing things. • 𝗛𝗮𝗿𝗱 𝗱𝗶𝘀𝗸 → 𝗩𝗲𝗰𝘁𝗼𝗿 𝗗𝗕 / 𝗹𝗼𝗻𝗴-𝘁𝗲𝗿𝗺 𝘀𝘁𝗼𝗿𝗮𝗴𝗲. Large capacity, but slow to access. You retrieve from it, not compute in it. • 𝗗𝗲𝘃𝗶𝗰𝗲 𝗱𝗿𝗶𝘃𝗲𝗿𝘀 → 𝗧𝗼𝗼𝗹 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻𝘀. The interfaces that let the model interact with the outside world. Code execution, web search, file I/O. • 𝗢𝗽𝗲𝗿𝗮𝘁𝗶𝗻𝗴 𝘀𝘆𝘀𝘁𝗲𝗺 → 𝗔𝗴𝗲𝗻𝘁 𝗵𝗮𝗿𝗻𝗲𝘀𝘀. This is the key layer. It manages everything: which tools to call, what fits in memory, when to retrieve, how to recover from errors, and when to stop. And then there's the 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 layer. That's the "agent" itself. Not a piece of software you install, but emergent behavior that arises when the OS does its job well. This is why two products using the exact same model can perform completely differently. LangChain changed only their harness infrastructure (same model, same weights) and jumped from outside the top 30 to rank 5 on TerminalBench 2.0. The model didn't improve. The operating system around it did. The article below is a deep dive on agent harness engineering, covering the orchestration loop, tools, memory, context management, and everything else that transforms a stateless LLM into a capable agent.