Sharing AI tools and insights to help you make money。分享AI工具和见解来帮助你赚钱。
chengdu
Joined March 2009
- Tweets4.8K
- Following641
- Followers266
- Likes14K
清风徐来028 retweeted
Come build your next app or agent with us in the API.
Eligible ChatGPT Pro and Business subscribers can claim up to $500 per month in API credits.
Get started: chatgpt.com/settings/billing
清风徐来028 retweeted
刚刚评论区和别人杠上了，硬说gpt赠送的api额度没法在codex里用。
怎么就不能用了？
一共三步：
1. 打开platform.openai.com/api-keys 新增api key
2. 在cc-switch（Magpie同理）新增一个供应商，填入openai的接口基地址和api key：api.openai.com/v1
然后选择需要使用的模型，保存。
3. 切换到这个供应商，重启codex
如果有动手能力的也可以直接改config.toml
清风徐来028 retweeted
Everything I learned building a full game with GPT 5.6 Astra, start to finish. Long read.
清风徐来028 retweeted
GPT-6 Astra + Blender + Godot Engine
GPT-6 仅用 20X 3% 的额度，帮我做出了一个完整的 3D Roguelike 游戏关卡
昼夜循环、天气系统、3D 建模、贴图、武器切换、技能系统，以及近战和远程的小怪
清风徐来028 retweeted
Given some of the results I'm seeing recently, it's pretty clear Codex is a good harness.
But it will seem primitive in 2-3 months and we're about to go through another major evolution in how we use AI at the frontier. The next generation of models need more than your laptop.
清风徐来028 retweeted
OpenAI 公布了 GPT-5.6 Sol 的最新提示词指南，核心可能就一句话：少写点废话，把任务边界讲清楚。
这份文档真的很值得看，它并不是教你继续堆更长的 system prompt，而是建议把 prompt 做轻：目标是什么、成功标准是什么、哪些证据必须有、什么时候该停、哪些动作需要确认。
这其实很符合新一代 Agent 的变化。
模型越来越强以后，过度规定每一步流程，反而可能拖慢它、增加 token、制造冲突。真正重要的是给它一个清楚的任务契约，让它在安全边界内自己选择路径。
我觉得这对 AI coding 特别有启发。
以后写 AGENTS.md / system prompt，重点不再是把模型管得越死越好，最重要的是把验收标准、权限边界和验证流程写得足够清楚。
developers.openai.com/api/do…
清风徐来028 retweeted
必须强烈推荐一下@canghe苍何老师的GPT2-image提示词网站（图片+提示词都有）：gpt-image2.canghe.ai/#galler…
太有想象力了！全网所有爆火的案例都收集齐了！看一圈下来，感觉脑子里面又多了很多好玩的想法，创造力就该在这样的环境里面成长。
但是不知道能不能提个小建议：现在只能复制prompt，不能直接在网站上预览完整的prompt，些许不方便。希望增加一个点击一下直接在网站里面预览的功能😋
清风徐来028 retweeted
We're delighted to announce that MiniMax M2.7 is now officially open source.
With SOTA performance in SWE-Pro (56.22%) and Terminal Bench 2 (57.0%).
You can find it on Hugging Face now. Enjoy!🤗
huggingface：huggingface.co/MiniMaxAI/Min…
Blog: minimax.io/news/minimax-m27-…
MiniMax API: platform.minimax.io/
Readers added context they thought people might want to know
The MiniMax M2.7 model weights are publicly available, but under a license prohibiting commercial use without authorization. This does not meet the Open Source Initiative's definition of open source, which requires allowing commercial use.
License: huggingface.co/MiniMaxAI/Mini…
OSI: opensource.org/osd
清风徐来028 retweeted
阿里巴巴官宣认领HappyHorse-1.0！
开放 API ，不开源模型
@HappyHorseATH 在 Artificial Analysis Video Arena 的所有排行榜上位居 #1 或 #2。
在我们“不带音频”的排行榜上，HappyHorse-1.0 稳居第一。在我们的“带音频”排行榜上，其 Elo 分数几乎与字节跳动的 Dreamina Seedance 2.0 相同。
HappyHorse-1.0 由 @AlibabaGroup 创建，它支持所有四种视频生成模式：文本到视频和图像到视频，每种模式均支持带和不带原生音频。API 访问计划于 4 月 30 日推出。
Revealing HappyHorse-1.0 as the latest video model from Alibaba!
@HappyHorseATH has landed in #1 or #2 across all of the leaderboards in the Artificial Analysis Video Arena.
In our ‘without audio’ leaderboards, HappyHorse-1.0 is comfortably in first place. In our ‘with audio’ leaderboards, its Elo score is almost identical to ByteDance’s Dreamina Seedance 2.0.
HappyHorse-1.0 was created by @AlibabaGroup, and it supports all four video generation modalities: Text to Video and Image to Video, each with and without native audio. API access is planned for launch on April 30.
See below for example generations of HappyHorse-1.0 in the Artificial Analysis Video Arena 🧵
清风徐来028 retweeted
🔥兄弟们！最近给小龙虾（OpenClaw）装上 XCrawl 后，数据采集直接起飞了！
今天就赶紧给你的龙虾装上这个铁钳吧！关键很省积分！
以前靠 jina. ai 那种轻量 Reader，抓抓简单页面还凑合，一碰到反爬强或者 JS 动态页就抓瞎，解析得乱七八糟😂
现在 XCrawl 直接给它补上四大 Skill，这下让我的龙虾从手变成了“铁钳”啊！
• scrape 单页精准抓
• crawl 全站异步爬
• map 站点结构探测
• search 多引擎 SERP
输出干净 Markdown + 结构化 JSON，还带智能提取，内置住宅代理+指纹，成功率稳很多，几乎零清洗就能直接喂 Agent。
✍🏻配置方法：注册拿个 API Key，git clone 一下 xcrawl-skills 仓库，把 4 个 Skill 文件夹复制到 OpenClaw 的 skills 目录，Key 丢进 config.json，重载一下就完事了，整个过程 5 分钟不到！
（👻邪修方式就是把文档key丢给OpenClaw 就可以自己完成非常容易）
我现在每天都这么干：让 Agent 用 xcrawl-search + scrape 扫全网 AI 科技新闻，一键输出结构化摘要和亮点，10 分钟就整理出当天最值得写的 𝕏 推文素材和选题！
以前手动刷半天，现在效率直接起飞～
想玩的兄弟直接注册领 1000 积分，30秒即可开始，无需绑定卡！
🤝 Paid partnership
清风徐来028 retweeted
Try the latest @Grok Imagine.
It only gets better from here.
Grok Imagine Video continues to lead at #1 across both Text to Video and Image to Video in the Artificial Analysis Arena. See how Grok Imagine Video compares to Kling 2.5 Turbo, Veo 3.1 Fast Preview, and Vidu Q3 Pro below
Grok Imagine Video is now also live in our Video with Audio Arena. You can compare and vote on it alongside Veo 3.1, Sora 2, and Kling 2.6 Pro.
Built by @xAI (led by @elonmusk), the model is available via the Grok Imagine API and is priced at $4.20/min with audio, making it cheaper than Veo 3.1 Preview ($12/min) and Vidu Q3 Pro ($9.60/min), while remaining competitive with Kling 2.6 Pro.
Here are example Image to Video generations from the Arena 🧵
Prompt [1/5]: Push in and cut to a frontal close up shot once the archer fires the arrow.
清风徐来028 retweeted
Grok Imagine only gets better from here
xAI's Grok Imagine takes the #1 spot in both Text to Video and Image to Video in the Artificial Analysis Video Arena, surpassing Runway Gen-4.5, Kling 2.5 Turbo, and Veo 3.1!
Grok Imagine is the latest video model from @xAI, and joins an increasing roster of models such as Google's Veo and OpenAI's Sora supporting native audio generation alongside video output.
The model is available via xAI's new Grok Imagine API only. It is priced at $4.20 per minute including audio, matching Kling 2.5 Turbo (without audio) and cheaper than Veo 3.1 Preview ($12/min) and Sora 2 Pro ($30/min).
Congratulations @elonmusk and the xAI team.
See below for comparisons between Grok Imagine and other leading models in our Artificial Analysis Video Arena
清风徐来028 retweeted
他在 66 天内完成了传统团队数月的工作量。他说：“如果你能驾驭这些工具，你现在的产出速度能媲美一年前的一家公司。”
看高手用AI后会多猛，比我们这类普通小白强太多了
清风徐来028 retweeted
We have a lot of exciting launches related to Codex coming over the next month, starting next week. We hope you will be delighted.
We are going to reach the Cybersecurity High level on our preparedness framework soon. We have been getting ready for this.
Cybersecurity is tricky and inherently dual-use; we believe the best thing for the world is for security issues to get patched quickly. We will start with product restrictions, like attempting to block people using our coding models to commit cybercrime (eg ‘hack into this bank and steal the money’).
Long-term and as we can support it with evidence, we plan to move to defensive acceleration—helping people patch bugs—as the primary mitigation.
It is very important the world adopts these tools quickly to make software more secure. There will be many very capable models in the world soon.