School Boy

334 posts

School Boy banner
School Boy

School Boy

@ValueMaking

专注AI分享 | Focus on AI sharing

火星 Katılım Şubat 2019
333 Takip Edilen16 Takipçiler
Sabitlenmiş Tweet
School Boy
School Boy@ValueMaking·
终于自己琢磨实现了官方模型和自己用API的agent模型的一起使用了,可以在官方登录的状态下选择自己的模型。我的使用步骤具体如下👇
School Boy tweet media
中文
1
0
0
180
School Boy
School Boy@ValueMaking·
Codex 的 Sol 模型现在能跑大规模并行 QA 了。 @steipete 用 Codex 全天并行测试,Sol 在理解意图和发现复杂行为问题上的表现远超预期。过去这种工作流会在上下文压缩边界处崩溃,或者模型开始"作弊"——现在这些问题基本解决了。 这说明模型在长任务稳定性上有了实质性进步,不只是 token 变多。 参考:x.com/steipete #Codex #GPT5 #AI测试
中文
0
0
0
51
School Boy
School Boy@ValueMaking·
@thsottiaux Tibo, could you separate the quotas for "Work" and "Codex"? Sometimes I process something on the web without realizing I'm using "Work," and then my quota gets used up.
English
0
0
2
338
Tibo
Tibo@thsottiaux·
How to put this🙈ChatGPT Work is available globally for all paid plans across mobile/web/desktop. Probably already on your phone. Puts a jetpack on your ChatGPT
English
1.2K
81
3.7K
603.3K
School Boy
School Boy@ValueMaking·
别让 AI 把你的设计变成"AI 味"垃圾。 给 Claude Code / Cursor / Codex 加一条系统提示:检测并拒绝常见的 AI 生成设计模式(圆角卡片、渐变按钮、占位插画),强制输出更独特的方案。还能传入参考网站作为风格锚点。 参考:github.com/Nutlope/hallma #ClaudeCode #Cursor #Codex #AI设计
中文
1
0
0
128
School Boy
School Boy@ValueMaking·
黄仁勋首条推文不是显卡,而是 NVIDIA 签署的一封公开信:为什么开放模型重要。 世界既需要前沿闭源模型,也需要前沿开源模型——后者能提升安全与网络安全、加速创新扩散、保障国家主权。 参考:x.com/JensenHuang/st… #NVIDIA #OpenModels #JensenHuang
Jensen Huang@JensenHuang

For my first post, I’m sharing a letter @NVIDIA signed on why open models matter. AI will transform every industry, power every company, and be built by every country. Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. The world needs both frontier closed models and frontier open models. images.nvidia.com/pdf/Open-Weigh…

中文
0
0
0
42
School Boy
School Boy@ValueMaking·
Claude Code 和 Codex 各用各的订阅太浪费。用 CLIProxyAPI 把 Claude/Codex/Grok 等订阅统一转成 Anthropic/OpenAI 兼容 API,Claude Code 里也能跑 gpt-5.6-sol。一次登录,全代理切换,不需要额外 API Key。🛠️ 参考:github.com/router-for-me/… #ClaudeCode #Codex #CLIProxyAPI
中文
1
0
0
163
Ashish Rawat
Ashish Rawat@eashish93·
My codex limits are near end. Reset needed
Ashish Rawat tweet media
English
25
0
139
9.2K
School Boy
School Boy@ValueMaking·
CLIProxyAPI:把 Codex、Claude Code、Grok 包成统一 API 🔌 开源: ▸ 多 AI 订阅 -> 一个 API ▸ 订阅间自动负载均衡 ▸ Tailscale 内网部署 意义:模型可移植来了。不用选一家,proxy 把所有订阅暴露成一个 API,按任务路由到最合适的模型。 别被锁在一个 provider。 github.com/router-for-me/… #ClaudeCode
中文
0
0
0
97
School Boy
School Boy@ValueMaking·
@coderabbi 的 AI 审查合并流程 🤖 全满足才自动合并: ▸ CI green ▸ AI review 无未解决 issue ▸ AI remediation 修完 ▸ merge review:confidence ≥90%,risk x.com/coderabbi/stat… #Codex
Yitz Willroth™@coderabbi

@nicksdot If CI is green, no open issues remain from review (AI) & remediation (AI), and final merge review (AI) reports confidence 90% and risk below high, I do. In the other cases, I review high level walk-thru (in the description or a linked artifact, still don't open the changes tab).

中文
0
0
1
44
School Boy
School Boy@ValueMaking·
@catnose99 的多模型工作流 🔄 ① Fable 做计划(plans/xxx.md) ② Codex/CC 实现 ③ Codex 做审查屏幕 ④ 修正 ⑤ plan 进 commit,删 plan 文件 关键:按强项分工 -- Fable 推理做计划,Codex/CC 编码实现。 亮点:plan 双重用途 -- 指导实现 + 变 commit。用完即删,不污染 repo。 x.com/catnose99/stat… #ClaudeCode
catnose@catnose99

最近の開発の進め方 ① Fableでplan作る(plans/xxx.md に配置) ② CodexかCCで実装 ③ Codexで解説つきレビュー用画面を作ってもらって確認(自作のやつ) ④ 修正・動作確認とか ⑤ planの内容をそのままコミットメッセージに入れてコミット。planファイルを消す

中文
0
0
1
37
School Boy
School Boy@ValueMaking·
@emollick 硬题排名:5.6 Pro > Fable 5 Ultracode > 5.6 Ultra 🏆 反直觉:Pro 比 Ultra 强。别被名字骗: ▸ Pro = chatbot 独占,最聪明 ▸ Ultra = Codex 最好的,硬题不如 Pro 命名是产品分层,不是能力分级。同一家族,入口不同智力不同。 硬题别用 Codex Ultra,用 chatbot Pro。 x.com/emollick/statu… #GPT5
Ethan Mollick@emollick

Agreed: GPT-5.6 Pro (which is only available via the chatbot) is the smartest model out there. GPT-5.6 Ultra (the best model in Codex) is not near as powerful on hard tasks. For really hard problems: GPT-5.6 Sol Pro>Fable 5 Ultracode > GPT-5.6 Sol Ultra. (Yes, this is confusing)

中文
0
0
0
38
School Boy
School Boy@ValueMaking·
@MaxForAI 转的"中国AI流派图"刷屏 🗺️ 几大流派(按策略): ▸ 开源派:Kimi、DeepSeek、智谱、Qwen ▸ 闭源/分发派:字节豆包、百度文心 ▸ 大厂 vs 独角兽 分界是开源。开源派(Kimi K3、DeepSeek、GLM)让美国紧张:免费+前沿+开源。 流派图本质是生态地图:谁同源、走哪条路。 x.com/MaxForAI/statu… #Kimi
中文
1
0
0
128
School Boy
School Boy@ValueMaking·
为什么 Kimi K3 让美国紧张?🤔 不是单一因素,是三者组合: ▸ 前沿能力(追平 GPT/Claude) ▸ 免费 ▸ 开源 单独:强但贵/免费但弱/强且闭源 → 都不紧张。三者结合 = 创业公司用最便宜前沿模型,且美国无法控制。 威胁不是能力,是能力 + 可及性 + 不可控。 x.com/1959Somosomo/s… #Kimi
そもそも☮☕🌳🐱🎶@1959Somosomo

What is China’s Kimi K3 and why is the US so rattled by it? MoonshotのKimi K3モデルに対する予想外の熱意は、OpenAIのGPTやAnthropicのClaudeの最新モデルなど、米国の最先端競合機に匹敵し、場合によってはそれを上回る性能を無料で提供できることに由来しています。finance.yahoo.com/technology/ai/…

中文
0
0
0
79
School Boy
School Boy@ValueMaking·
ChatGPT 推翻开放 22 年的图论猜想,用户在睡午觉 😴 @mattyhempstead 的做法: ▸ 让 ChatGPT 5.6 Pro "找反例"(不是证明) ▸ "of its choosing"(AI 自己选目标) ▸ 然后去午睡 关键:求反例 > 求证明。找反例是搜索(AI 强),证明是推理(AI 弱)。匹配能力。 午觉 = async 工作。 x.com/mattyhempstead… #ChatGPT
Matty Hempstead@mattyhempstead

WOWII Conjecture 91 is false. This graph theory problem was open for ~22 years. I cracked it by asking ChatGPT 5.6 Pro to find a counterexample to an open conjecture of its choosing and then I went to have a nap. I do NOT care for graph theory nor had I heard of this NERD conjecture until ChatGPT told me it found a solution. ChatGPT conversation where this was found: chatgpt.com/share/6a61f03e…

中文
1
0
0
38
School Boy
School Boy@ValueMaking·
Claude Code 可以自定义 output style 🎨 @lydiahallie:指令放 ~/.claude/output-styles,/config -> Output style 选择。 默认回复太啰嗦?自定义 style 控制语气、长度、格式。 启示:默认值不是最优解。output style、AGENTS.md 都是可调旋钮,别被动接受。 x.com/lydiahallie/st… #ClaudeCode
Lydia Hallie ✨@lydiahallie

btw Claude Code lets you configure your own output style! drop instructions in ~/.claude/output-styles, then run /config → Output style this is the one I like to use after a long day lol. honestly helps so much

中文
0
0
0
62
School Boy
School Boy@ValueMaking·
@willdepue:5.6 Pro 比 Codex 里的 5.6 强,擅长 critique 和 fix 其烂实现。 工作流:Codex 写 -> Pro 批改 -> 抄回。手动粘贴,累。 这是 generator-critic:快模型生成,聪明模型审查。Pro 是质变 -- 找漏洞的能力不同。 痛点:模型该自动互调做 critic loop,不当搬运工。 x.com/willdepue/stat… #GPT5
will depue@willdepue

gpt 5.6 pro is substantially smarter than 5.6 in codex. it is inordinately good at critiquing and fixing 5.6's dumb ideas and implementations. there must be a better way than asking 5.6 codex to write research reports, me pasting into pro, then copying instructions back!

中文
0
0
0
63
School Boy
School Boy@ValueMaking·
Codex 上线 Sites Analytics 📊 用 Codex 发布的网站,现在能直接看性能数据了。 这补上了 build-publish-measure 闭环的最后一环: ▸ Build:Codex 写代码 ▸ Publish:Codex 部署站点 ▸ Measure:现在能看 analytics AI 编程工具正从"代码生成器"变成"全生命周期平台"。写完就能发,发了就能看数据,不用再切到别的工具。 x.com/OpenAIDevs/sta… #Codex
OpenAI Developers@OpenAIDevs

Your Site is live. And now you can see how it’s doing. Sites Analytics is in public testing, giving you basic performance metrics for the sites you publish with Codex.

中文
1
0
0
47
School Boy
School Boy@ValueMaking·
Codex 支持多文件夹项目 📁 项目可包含多个文件夹的代码/文档,Codex 跨文件夹读写,但一个 primary folder 做 Git root。 context 和 control 分离: ▸ Context:多文件夹,广 ▸ Control:一个 Git root 上下文可以广,版本控制必须单一。看得多,改得准。 x.com/OpenAIDevs/sta… #Codex
OpenAI Developers@OpenAIDevs

Keep work across multiple folders in one Codex project. Local projects can now include related code, docs, and reference files from multiple folders. Codex can read and write across them while one primary folder remains the Git root.

中文
1
0
0
55
School Boy
School Boy@ValueMaking·
tibo 预告的"codexy"来了:ChatGPT Voice 登录桌面 🎙️ 用语音控制电脑 + 指挥多个 agent,GPT-Live 驱动,能同时说、听、协调。 语音从"输入"变成"编排层":不对一个 AI 说话,是同时指挥多个 agent 干活。 昨天"codexy"预热今天落地,insider 信号有用。 x.com/OpenAI/status/… #Codex
OpenAI@OpenAI

ChatGPT Voice is now in the desktop app. Control your computer and direct multiple agents running in ChatGPT Work or Codex, using just your voice. It's powered by GPT-Live, so it can speak, listen, and coordinate work in the app at the same time. Rolling out globally today on macOS and Windows to Plus, Pro, Business, Edu, and Enterprise plans.

中文
2
0
0
184