OpenAI 刚刚发布了 Codex AI 代理在你喝咖啡的同时，就能在云端生成可用于发布的代码。🤯 ChatGPT 的专业版、企业版和团队版用户将陆续开放使用。 https://t.co/J4QlQZauV8

OpenAI发布Codex👏： 00:00 介绍 Codex 01:38 设置并启动任务 05:05 环境配置 07:20 使用 Markdown 文件引导 Codex 12:00 Codex 的起源故事 14:25 验证和审查 Codex 执行的任务 17:28 在 OpenAI 使用 Codex 20:48 通过 ChatGPT 提供使用 21:22 路线图和 Codex CLI 更新 https://t.co/VxtZdnCTr0

OpenAI 在最近的版本更新中，增加了对购物功能的支持和优化（目前已向所有可使用 ChatGPT 的地区的 Plus、Pro、免费及未登录用户开放）。 AI的尽头也是带货了。

OpenAI CPO Kevin Weil says it's inevitable we'll let AI agents spend on our behalf, but with strict limits. “I want this agent to buy this thing for me, but only spend up to $50, and only at a grocery store.” He sees smart contracts as the control layer. Programmability is how we align agent behavior with human intent.

OPEN SOCIETY ADMITS: YES, WE FUND PROTEST GROUPS Protests targeting Elon, Tesla, Trump, and Israel are labeled "grassroots"—but major Democratic donors are footing the bill. Groups like Jewish Voice for Peace and IfNotNow, leading anti-Israel protests, are bankrolled by foundations seeded by Soros, Rockefeller, and the Pritzkers. The Open Society Foundation admits it funds groups that may organize protests, while denying direct coordination. The pattern is clear: if there’s a left-wing protest, there’s donor money somewhere making sure it happens. Source: WOAI, OSF

OpenAI 发布了GPT-4.1 提示工程指南 以前许多典型的最佳实践和提示依旧适用于 GPT-4.1 由于GPT-4.1 更严格、更字面地遵循指令，会非常严格地按照字面指令去执行任务。 这使得它对明确、清晰的提示尤其敏感。也就是说，只要你发现 GPT-4.1 的表现与预期不符，通常只需增加一句简洁明确的说明，就能迅速把模型引导到正确的行为上。 过去的模型（如 GPT-4） 会更自由地揣测或推断用户指令和系统提示背后的真实意图，即使提示不够精确，也可能猜出用户的意图并完成任务。 所以开发者需要对原有的提示方式进行一定调整（迁移）才能使用。 官方给出了详细的针对 GPT-4.1 的提示工程（Prompting）最佳实践，从基础原则到高级策略，帮助开发者高效构建提示以提升模型表现。

OpenAI 正式发布 GPT-4.1 系列：百万 Token 长文本、更强编码、价格更优！ OpenAI 隆重推出专为开发者打造的全新模型系列——GPT-4.1！该系列包含三个强大的新成员：旗舰级的 GPT-4.1、高速的 GPT-4.1 Mini，以及有史以来最小、最快、最经济的 GPT-4.1 Nano。这些模型在编码能力、复杂指令遵循、长文本处理（首次支持高达一百万 Token 上下文，且无额外费用）以及多模态理解方面均实现了显著提升，性能全面超越 GPT-4o，并在关键指标上比肩甚至超越 GPT-4.5。视频中，OpenAI 团队成员详细介绍了这些模型的性能优势、基准测试结果，并通过精彩的实时演示（如现场编写功能完善的前端应用、处理超长日志文件）展现了它们的实际能力。同时，他们还分享了更优惠的定价策略，并宣布了 GPT-4.5 的后续计划。快来了解这些为开发者量身定制、现已在 API 中可用的新一代 AI 模型吧！

OpenLora is the designated deployment engine for specialized models in the OpenLedger ecosystem. By leveraging just-in-time adapter switching, OpenLora enables the efficient serving of thousands of fine-tuned LoRA adapters on a single GPU, drastically reducing deployment costs. Unlike generic models, OpenLora-powered specialized models require fewer input tokens and produce more optimized, precise outputs - minimizing both input and output token size while delivering task-specific performance at scale. Faster, Smarter, and Specialized.

OpenLedger Epoch 2 testnet nodes are now live! Epoch 1 node runners will receive priority access for node running, followed by new signups. Initially, data contribution will be available for whitelisted users only, with broader access (including epoch 1 node runners) coming later. At the moment, only the Chrome Extension nodes are live, with PC nodes coming soon. Register https://t.co/AYY21Ho9x0 Download Chrome Extension and keep running to earn points: https://t.co/7VqSHZ8t83

OPEN ORDER💖 💖Top up Love and Deepspace Event: “Caleb’s Myth: Fallen Cosmos” Bundling Packs (Harga lebih murah): • [15 Wish + 150 Dias] Pack I (5x) + Pack II (5x): Rp 105.000 • [21 Wish + 150 Dias] ] Pack I (5x) + Pack II (5x) + Pack III (1x): Rp 170.000 • [45 Wish + 150 Dias] Pack I (5x) + Pack II (5x) + Pack III (5x): Rp 430.000 • [95 Wish + 150 Dias] Pack I (5x) + Pack II (5x) + Pack III (5x) + Pack IV (5x): Rp 1.090.000 • [200 Wish] Pack VI (10x): Rp 3.550.000 Also available: ☑️ NEW Limited Wish Packs (Via login Inf*ld) ☑️ Secret Promise/Heartfelt Vow (Via login Inf*ld) ☑️ Aurum Pack 7 Days (Via login Inf*ld) ✅ 100% legal mendapatkan invoice resmi ✅ proses fast, 10-60 menit ✅ terpercaya sejak 2020 dengan 5k+ testimoni ‼️ Order? DM/WhatsApp: https://t.co/83zntke4P7 ••• #LoveandDeepspace# #LADS# #LND# #LAD# #warunglad# zayne xavier sylus rafayel mc li

. @OpenledgerHQ x @eigenlayer: The full episode Our core contributor @Ramkumartweet sat down with @sreeramkannan (founder @eigenlayer) to discuss verifiable AI, crypto's app problem, AI's trust problem and much more Check out the video below to learn more Timestamps: 00:00 Sreeram's introduction 00:29 How to build verification for AI systems 01:27 The Crypto and AI problem 03:10 Unstoppable AGI 05:08 What is Verifiable AI 06:05 The Verifiability Problem with Centralized AI 07:15 The current state of AI and its future 08:48 How humans are getting better at intelligence extraction 10:31 What is OpenLedger doing 11:50 Where LLMs fall short and how co-ordination helps 12:50 SLMs are already there 13:29 Crypto as an incentivisation mechanism 15:38 Apps with commitments, achieving product market fit 16:50 Existing examples of commitments in society 18:24 Investable AI 19:00 Data contributors as co-owner 21:30 The solution for the incentivisation problem 22:25 Explaining explainable AI 23:58 Trusting the trustless way 25:07 Data is the new oil 25:40 The idea of openledger is very crypto native 26:50 David v Goliath (in the data perspective) 28:03 What Eigen is working on 29:06 Making unstoppable verifiable AI practically possible 32:20 What Sreeram uses for productivity 33:30 Make Eigen Labs agentic 34:16 (community question) How Sreeram frames his prompts 36:08 (community question) What's the momentum driver for blockchain apps and what is Sreeram most excited about? (ZkTLS) 38:54 (community question) Where is attribution and verification really needed in AI?

OpenLedger Testnet Epoch 2 is officially LIVE! @OpenledgerHQ Whitelisted members, access to these models is rolling out now in batches. The Data Intelligence Layer for node runners is also launching soon! Register here: https://t.co/qRjThoSH1B https://t.co/c3eOWAAZgf