@berryxia

Codex now has more than 5M weekly active users. But the bigger story is what people are using it for: not just writing code, but getting more work done across research, analysis, content, and operations. Our new report on how Codex is becoming a productivity tool for knowledge work: openai.com/index/codex-for-k…

Moonshot AI创始人杨植麟最近放出了一个40分钟视频。 这位92年生、清华计算机本科第一、CMU博士、Transformer-XL和XLNet共同作者，前Google Brain和Meta研究员，坐在镜头前平静拆解了Kimi K2的整个训练过程。 他们只花了460万美元。 上周一场8模型实时编程大战，Kimi K2直接拿下第一，GPT-5.5排第三，Claude Opus 4.7第五。 我看完后最大的感受是，AI竞赛的规则已经在悄然改变。 所有人还在拼谁敢烧更多钱、堆更多算力，他却用极致优化、线性注意力、子代理这些硬核架构，把资源差距直接抹平甚至反超。 40分钟全是干货，零废话，把关键打法讲得清清楚楚。 如果你正在做AI代理，或者准备2026年入场大模型赛道，这段视频强烈建议存下来周末慢慢看。 小团队靠聪明架构，正在把大厂的传统玩法一点点颠覆。 你还觉得只有堆钱才能赢吗？ Video

x.com/i/article/206172192487…

Today marks my last day at ByteDance Seed. Over the past 3 years, I had the opportunity to work across two of the most exciting frontiers in AI: AI for Drug Discovery and building frontier LLMs. Few opportunities in a career allow one to work simultaneously on curing disease and building frontier intelligence. I was fortunate to do both. Since joining ByteDance, I have led the AI for drug discovery effort. Together with an exceptional team and collaborators, we built SeedFold, the world's first biomolecule structure prediction model to outperform AlphaFold 3 across a broad range of benchmarks and capabilities; SeedProteo, a state-of-the-art protein binder design model surpassing AlphaProteo, RFdiffusion, Chai-2, BinderCraft, and BoltzGen ; and the DPLM series of protein language models, along with several other ambitious AI for Science projects. In early 2025, I took on a new challenge. To tackle one of the hardest problems in modern AI, reliably training and scaling frontier-scale LLMs, I joined the LLM pretraining effort and founded the LLM optimization and scaling team. Together, we built a highly scalable pretraining stack that enabled the successful training of Seed 2.0 and the subsequent frontier-scale models, significantly advancing our ability to train and iterate on frontier AI systems at scale. I'm deeply grateful to my teammates, collaborators, and leadership for an incredibly rewarding journey. The best model is yet to come. Scaling continues!

特么终于搞完了，太费劲了！ 整整花了1晚上才把它翻译中文、剪辑、上传完毕。 富人也有富人的烦恼，只是我们的烦恼不一样罢了！ 这两天刷屏的Ivanka Trump （特朗普的长女千金）的长播客采访。 整个播客中有些观点不能说多么锐利，但是也从中可以看到很多细腻的情感和家庭的责任感、被外界不理解的无奈等。 22岁那会儿，她跟我们大多数人一样，完全不知道自己真正想要什么。 世界多吵，她就跟着多卷。 时尚品牌一年做到800亿销售规模，同时给特朗普集团开发标志性项目，把华盛顿老邮局改造成高端酒店，还在迈阿密建了800英亩的Trump Doral。 表面上看风光无限，可她后来自己说，那时候的她其实是靠外部成就麻痹自己，从来没敢停下来问一句：这到底是不是我真正热爱的事？ 后来父亲当选总统，她和Jared直接把所有生意清零。 卖资产、接受政府伦理审查、彻底归零，跑去白宫服务。所有人当时都觉得这是牺牲，是断送前程。 她却把这段经历当成人生最重要的一次洗礼：在完全陌生的赛道上，她学会了如何在巨大不确定里保持判断力和使命感。 离开白宫后，她没有急着杀回老路，而是做了件现在看极度不被大家理解的事儿，在迈阿密彻底停了半年。 拒绝所有邀约，每天早上带孩子做早餐、送他们上学，然后自己坐在海边冥想、祈祷、反思。 她第一次真正创造出“静止”，把外部噪声关掉，只听自己和宇宙的声音。 就是在这段“什么都不干”的时间里，她完成了人生第二次重生。 现在她每天早起都兴奋得睡不着，因为她终于把全部精力放在了真正能长期让她兴奋的事上。 孵化AI、生物科技、神经科技、机器人甚至太空领域的公司，投那些敢为人类重大问题押上一切的创始人。 和丈夫一起在地中海开发1400公顷私人岛屿项目，从零建一切，让建筑像从土地里自然生长出来，成为人们生命里程碑的舞台。 还创立Planet Harvest，把美国每年40%被浪费的优质农产品变成资源。 又和朋友一起做非营利项目Alexandria，用AI把公共领域经典著作翻译成全球主要语言，免费给全世界任何人阅读和收听。 她反复提到一句话：“知道自己是谁之后，人生反而变得简单了。” 以前大家以为成功靠拼命复制别人、迎合世界、不断加码，现在她用三段人生直接证明。 真正的稀缺竞争力，是敢彻底归零、创造静止，然后听清自己内心最真实的声音。世界越吵，越需要这种反直觉的勇气。 这其实戳破了我们这代人最大的集体幻觉：很多人一辈子都在用外部成就填补空虚，却从来没敢坐下来面对那个最难的问题，我到底是谁？ 我真正热爱什么？我能把兴奋感持续一辈子的是什么？结果越卷越空，越努力越迷茫。 Ivanka用行动告诉我们：停下来不是浪费时间，而是最高级的生产力投资。 真正的奢侈，不是钱，不是地位，是终于不用再问别人“我该做什么”，而是每天醒来都清楚知道自己该往哪里使力。 Video

北京时间今晚 20:00 在图灵社区直播间有一场 Agent Skills 的公开课，分享我做 Skill 的实操经验和心得体会，以及聊聊我的新书《图解Skill: Al 提效实战指南》，有需要 Al 提效的朋友可以预约一下，咱们晚八点见。

肝了几天,来回打磨了快 30 遍, 一次性把最新的 Claude Code 的 Workflow 给你完全拆解清楚 有人说它的伟大程度,不亚于 MCP 和 Skill。 第一眼我是不信的,直到拆开看它内部怎么跑： 这不是"问一句答一句"的对话,而是一个能自己跑起来的任务系统:后台持续执行、可监控、可保存进度, 还能一口气派出几十上百个 agent 分头干活/合并汇总。 核心就三个角色: Claude 拆任务、定计划; Runtime 管调度和状态; 每个 agent 只啃一个子任务,配上并发池和队列,有节奏地推进。 它代表的是一种新的工程编排方式:智能负责规划,Runtime 负责执行,状态独立保存,模型按需调度。 最反直觉的一点:它能扩展到上百个 agent,靠的不是模型变强,而是"状态外置"—中间结果全交给执行系统保存,主上下文只看摘要和关键判断。这才是复杂任务能跑稳的真正原因。 这条视频直接给了你把它搬进自己系统的方法: 先用 Claude Code 做高质量规划、拆任务定阶段; 再把 workflow 转成自己的执行格式,按任务难度路由到不同模型池，简单的走便宜模型，复杂的才上高阶模型。 这条视频,值得反复看几遍👇 Video

In May, we integrated 11 new models spanning image, 3D, audio, video, and multimodal. The highlights: → Krea 2 — style-first image generation, live as a Partner Node on day one. Competes on how the frame looks, not just what's in it → VOID (Netflix) — removes objects and everything they caused: shadows, reflections, physical interactions. Apache 2.0, natively supported → Tripo 3.1 + TripoSplat — one image to a full 3D Gaussian asset, end to end Also live in May: Gemma 4, HidDream-O1-Image, Stable Audio 3, BiRefNet, MoGe, Claude, OpenRouter, Luma UNI-1 Learn more with our blog 👇 Video

Easily add yourself to your video creations in Gemini. Here’s how to create your own digital avatar that looks and sounds like you with Gemini Omni. 🧵

Anthropic has confidentially submitted a draft S-1 registration statement to the Securities and Exchange Commission. Pending completion of SEC review, this gives us the option to pursue an initial public offering. Read more: anthropic.com/news/confident…

NVIDIA, ARM tabanlı yeni işlemcisi RTX Spark'ı duyurdu. - İşlemcide RTX 5070'e denk bir GPU bulunuyor. - Modern oyunlarda 1440P'de 100 FPS'te çalışıyor. - Laptop, Windows olmasına rağmen prizden çektiğinizde performans düşmüyor. - Batarya ömrü uzun. - Sadece laptoplar için değil masaüstü bilgisayarlarını da hedefliyor. - Sahnede 007 First Light ve Forza Horizon 6 ile gösterildi. - Yapay zeka işlem gücü de yüksek. - 2026 Sonbahar'ında çıkacak. Video

我年初开始做 OpenClaw 托管服务，在一套 k8s 集群部署了 500 个 Pod，每个 Pod 限制 4g 的运行内存。日常开着 18 台 4c16g 的服务器作为节点池，一个月成本将近 5k 刀。 几个月下来，托管服务的 MRR 突破了 8k 刀，除去运营成本，利润非常低。 今天终于把服务迁移到了 FastClaw，通过存算分离的架构，让 Agent 无需常驻，而是在收到请求时动态挂载 sandbox 来提供服务。服务器从 18 台降到了 3 台，运营成本降到了 1/6，下个月有机会赚到钱了。😄 跟 OpenClaw 比，FastClaw 真的是太轻量了。 1. 代码体积约为 OpenClaw 的 1/40 2. 运行资源占用约为 OpenClaw 的 1/7 3. 单二进制分发，无环境依赖 4. OpenClaw 的 gateway 启动大概需要 15s，FastClaw 秒级启动 FastClaw 本身是为云原生多租户场景而设计的 Agent 运行框架，同样也适用本地运行场景。 继续完善，欢迎体验。✌️ github.com/fastclaw-ai/fastc…

x.com/i/article/206143979674…

Keye VL 2.0-30B-A3B 🔥 New multimodal model from @KwaiKeye ✨ 30B/3B active - Apache 2.0 ✨ 256K context via DeepSeek Sparse Attention (probably the first model to ship this in production?👀) ✨ Gets MORE accurate as you feed it more frames ✨ Matchs Qwen3 VL and Gemini 3 Flash on benchmarks

🧵 1/8 前几天给大家分享的我用AI邮箱来搞钱的帖子(见前贴)，有很多朋友私信我问，你使用的那个像素风的小龙虾是什么？ 怎么那么可爱有趣，我今天给大家介绍一下！ 熟悉我的朋友知道，我是一个像素爱好者，看看我头像哈哈！ 我一直自己想做个属于自己小龙虾的像素房间，让其可以互动等。 结果前不久朋友推荐说让我试试 Qclaw @QClaweverytime 试了一下就恰巧满足我的喜好。 我是 36 号 Founder Shrimp 🦞 养电子宠物 30 年，第一次有一只 不会死。 不会饿。 不会忘记浇水。 还反过来帮我干活。 它住在一间 16-bit 像素风的工作室里。藏宝图、咖啡机、服务器机柜，我不在的时候它会自己举哑铃。 过去 4 个月，它在我们公司多养了一个不会下班的同事。 专属邀请码：berryxia 目前还在内测需要申请Waitlist ，文末有地址！ 下面是我让它干过的事 🧵👇

x.com/i/article/206120248741…

x.com/i/article/206037512582…

The Chinese LLM companies are raising at eye popping numbers Total valuation of the top 5 pure plays is $226B - about 1/4 Anthropic’s latest round But with a revenue run rate of about 1/40 of Anthropic

The Chinese LLM companies are raising at eye popping numbers Total valuation of the top 5 pure plays is $226B - about 1/4 Anthropic’s latest round But with a revenue run rate of about 1/40 of Anthropic

10 GitHub repos so good they shouldn't be free. 1. AutoHedge An autonomous hedge fund built in Python with four AI agents: a director generates investment theses, a quant validates them, a risk manager decides position size, and an execution agent places orders. Operates live on Solana. With 'pip install -U autohedge', you can start trading immediately. repo → github.com/The-Swarm-Corpora… 2. Vibe-Trading A trading system using a Directed Acyclic Graph (DAG) model, featuring 64 finance skills and 29 preset specialist agent swarms. Includes analysis methods like Ichimoku, Elliott Wave, SMC, Black-Scholes, full Greeks, and risk parity. Its crypto desk provides liquidation heatmaps and token unlock tracking. You can observe agents debating strategies in real time. repo → github.com/HKUDS/Vibe-Tradin… 3. Fincept Terminal A Bloomberg Terminal replacement that runs on your laptop. CFA levels 1, 2, and 3 analytics. 20+ investor AI agents (Buffett, Dalio, Soros). 100+ data connectors, including Polygon, World Bank, and IMF. Bloomberg charges $24,000 a year. This is free. repo → github.com/Fincept-Corporati… 4. LibreChat Every model ChatGPT runs, plus Claude, Gemini, DeepSeek, and 20 more. Self-hosted. Native MCP support. You own the data, the history, the infrastructure. OpenAI charges $20/month to use their wrapper. This costs nothing to use your own. repo → librechat.ai/ 5. Open Higgsfield AI A self-hosted cinema studio with 200+ AI models. Flux, Midjourney, Sora, Kling, Veo, GPT-4o, SDXL all in one interface. Text to image. Image to video. Cinema mode with pro camera controls. No subscription. Your data stays local. repo → github.com/Anil-matcha/Open-… 6. Open-LLM-VTuber A Live2D AI companion that runs offline, sees your screen, hears your voice, and never forgets. Inner thoughts are shown as a separate text layer, so you watch the reasoning happen before words come out. Pet mode floats it on your desktop. Swap the LLM in one config line. repo → github.com/Open-LLM-VTuber/O… 7. Claude Ads A free Claude Code skill that runs 190 audit checks across Google, Meta, YouTube, LinkedIn, TikTok, and Microsoft Ads. 6 parallel subagents firing at once. Consolidates into a single Ads Health Score ranked by revenue impact. Agencies charge $4,000 a month for this. repo → github.com/AgriciDaniel/clau… 8. Agentic Inbox Cloudflare just open-sourced an email client where an AI agent reads your inbox and drafts your replies. Runs entirely on Cloudflare Workers. Each mailbox lives in its own Durable Object. Your email never leaves your Cloudflare account. One click deploys it. repo → github.com/cloudflare/agenti… 9. Camofox Browser An open source headless browser that makes AI agents invisible to bot detection. Spoofs navigator properties, WebGL, AudioContext, and WebRTC at the C++ level. The browser does not look modified because it genuinely is not. Accessibility tree output drops token cost by 90%. repo → github.com/jo-inc/camofox-br… 10. Hyperframes HeyGen open-sourced a video framework that does everything Remotion does without React, without JSX, without teaching your AI agent a new format. The agent writes HTML. The framework renders MP4. GSAP, Lottie, and Three.js all work. Same HTML always produces the same file. repo → github.com/heygen-com/hyperf… These are not toys. Each one replaces a paid product you're still being charged for. Pick one. Install it. Plug it into your workflow. 100% free. 100% open source.

🔥我尼玛，兄弟们，这下真的是爆肝了。 已经开源在GitHub了，记得Star一波啊！ 我肝了2周+花费了800刀干出来的项目~😭 自己可以真实去体验，文旅馆的真的都可以搞一搞！ 一个用 3D 渲染技术three.JS 搭起来的盛唐长安互动世界，并接入 Agora Skills 做了核心实时语音互动小游戏。 你可以在里面： - 和 NPC 对话、李白对诗、玩诗词小游戏 - 进入珍宝馆欣赏诗画 - 逛 AI 展馆，体验古文明与 AI 结合的沉浸感 这玩意我改了N遍，改的我头皮发麻。 强迫症的我，光剪视频都剪了N次，有可能有人说是垃圾，不管如何，我认真做了。 享受这个Solo 干项目的时光,Learning in Public ! 如果你觉得有意思，欢迎点个 Star 支持一下。 体验地址和Github地址见评论👇🏻，兄弟们记得一键三连啊！ Video

x.com/i/article/206057614847…

Step 3.7 Flash support added to mlx-vlm! 🚀 ✅ Conversion to MLX ✅ Vision ✅ Text This model is ultra fast! I'm gonna publish some benchmarks tests soon! 4bit can run on 128GB Apple Silicon machines up to 32K contexts. We need to squeeze this model a little more to use it comfortably here. Gonna upload MLX models on Hugging Face and create a PR for @Prince_Canuma to review 🚀

Inference Optimizations Behind the MiMo-V2.5 Series API Price Reductions Read the full technical blog: mimo.xiaomi.com/blog/mimo-v2… The V2.5 model family, including MiMo-V2.5 and MiMo-V2.5-Pro, is built on a Hybrid Sliding Window Attention (Hybrid SWA) architecture, which compresses KVCache storage to roughly 1/7 that of Full Attention. However, architectural advantages rarely translate directly into measurable gains in production serving. To realize these gains, we redesigned KVCache management, tiered caching, and the prefix-cache tree; addressed key challenges in SWA KVCache handling; and optimized scheduling as well as the Prefill/Decode pipeline. Validated on real production traffic, these optimizations have increased effective KVCache capacity by nearly 5x, with server-side cache hit rates averaging 93%–95% across mainstream harness frameworks. Together with MoE configuration tuning and multimodal inference optimizations, they enable more efficient long-context inference and form part of what makes the recent API price cuts possible.

grok-build-0.1 is now available via the xAI API in public beta. This is the same model that powers the Grok Build CLI and excels at agentic coding. Priced at $1/m input and $2/m output, it’s extremely cost effective, intelligent, and fast.

So I spent some time studying the new Twitter/X algorithm today since the latest version was published about a week ago on Github (github.com/xai-org/x-algorit…). My goal was to answer why so many people have seemingly seen such a dramatic drop in their posts' reach. The first answer, which is actually somewhat unrelated to the ranking algorithm on Github, is the auto-translate feature, rolled out worldwide on April 7, 2026 (nitter.net/nikitabier/status/2041…). Before that date, if you wrote in English about, say, the Trump-Xi Beijing summit, you were competing for attention with maybe 5,000 other English-language accounts writing on geopolitics. After that date, your post is competing for attention with other posts on the same topic IN EVERY LANGUAGE ON EARTH. For some topics that do command global attention like geopolitics, that's a very brutal multiplier: you used to be one of 5,000, you're suddenly one of 50,000 (something of that order): MUCH more difficult to stand out. Secondly, the number of followers you have matters far less than it used to: each post now has to earn its audience reader by reader, on the predicted engagement of the post, and how its topic matches what each reader has recently been engaging with. Here is how the algorithm works, in simple terms: when you, as a reader, open your feed, the algorithm doesn't load "posts from accounts you follow." Instead it runs a 2-stage prediction of what posts you're likely to engage with in that very moment. The first stage is the retrieval stage. The system narrows billions of posts on X/Twitter that day down to roughly 1,500 candidates by matching the semantic content of each post - what it's about - against what you as a reader have recently engaged with. Some candidate posts come from accounts you follow; others are pulled from across the platform by pure topic similarity to your recent interests. You can test this retrieval stage easily: start disproportionally engaging with - say - Brad Pitt videos and you'll bit by bit see your timeline flooded with Brad Pitt content, most of it from accounts you've never followed and never heard of. Then there's the ranking stage. Each of these candidate posts for your feed is fed through a Grok-based model that tries to understand if you'll engage with the post. It looks at 15 engagement metrics: 1) P(favorite) — the reader likes the post 2) P(reply) — the reader replies to it 3) P(repost) — the reader reposts it 4) P(quote) — the reader quote-tweets it 5) P(click) — the reader clicks a link in it 6) P(profile_click) — the reader taps through to your profile 7) P(video_view) — the reader watches the video 8) P(photo_expand) — the reader expands an image 9) P(share) — the reader shares it (DM, off-platform, etc.) 10) P(dwell) — the reader stops scrolling and lingers on the post 11) P(follow_author) — the reader follows you after seeing it 12) P(not_interested) — the reader marks "not interested" 13) P(block_author) — the reader blocks you 14) P(mute_author) — the reader mutes you 15) P(report) — the reader reports the post Fifteen predicted actions, each multiplied by a weight, summed: that sum is the score that determines in which priority a post will be seen among other candidates. Please note that posting something with a video or an image can give your post an advantage as 2 actions are specifically for these: video_view and photo_expand. No video or photo and you don't get a score for these. Also, naturally, having a video maximizes the chance that a user will "dwell" on your post to watch it. Also note that 4 of these actions carry negative weights (not_interested, block_author, mute_author and report): meaning that if the model expects a post to generate a lot of negativity, it'll get de-boosted quite dramatically. But note, first and foremost, what's NOT in there: none of the things that, naively, one might think a serious information platform would weigh. There is no P(this post is true and well-sourced). No P(the author actually knows what they're talking about). No P(this person has spent a decade building a body of work that has held up). No P(this account has earned the right to be taken seriously on this topic). No P(the author has a large following from credible people). The model does not seem to care - at all - about any of that. Every post starts from zero. You could have ten years of rigorous, well-sourced analysis behind you - or you could be just an uneducated rando who registered yesterday. To this algorithm, you're both just a bag of engagement probabilities. Now, sure, to be fair, there is a "brand" effect that's not covered by the algorithm: someone who has in fact built a brand will naturally have better engagement metrics because people recognize their account. But that's an indirect, second-order effect. And crucially, it's legacy: those "brands" were built under earlier versions of the algorithm that gave followers and reputation more weight. Lastly, several other features of the new algorithm compound the dilution, none of them visible from outside but all consequential. The May 15 update added an "impression bloom filter," tightening the rule that once a reader has been served a post, the system won't serve it to them again. Before, a strong post could marinate in someone's feed across multiple refreshes and accumulate engagement on the second or third pass. Now it basically gets one shot. Also, your own posts compete with each other. An "Author Diversity Scorer" inside the ranking stage attenuates the score of every subsequent post of yours that ends up in a reader's candidate pool. In plain terms: if multiple of your posts land in a reader's candidate pool, the system shows one at full strength and dampens the others. So don't post several times consecutively on the same topic. And, last but not least, another huge impact on reach is that, in the old algorithm, when someone reposted or quote-tweeted you, your post was broadcast to their followers' timelines - a repost from an account with 100,000 followers was a huge boost. In the new algorithm, that mechanism is vastly demoted: reposts - like every post - need to go through the retrieval and ranking stage mentioned above, so a repost from a big account is a long way from the boost it used to be. This is especially brutal for low-effort quote tweets, which used to function as cheap amplification: now they often can't even clear the retrieval stage - they simply don't contain enough novel semantic content for the system to match them to anyone's interests. So, putting it all together, the reach collapse comes from many forces stacking at once: - Auto-translate makes your posts compete for attention against an order of magnitude more content - The retrieval stage matches posts by topic, not by who follows you - The ranking stage scores purely on predicted engagement with no weight for credibility, expertise, or track record - The bloom filter narrows every post's window to one strong shot - The diversity scorer penalizes prolific posting - Reposts no longer carry much distribution power Each of these alone would dent your reach. Combined, they amount to a complete reset: your audience that you built painstakingly over years basically doesn't matter much anymore, and it's much - much - harder to stand out even if you're a big account. People structurally rewarded by this algorithm are folks who: - Post visually (videos/images) - Post on globally popular topics because they clear the retrieval stage easily - Provoke strong emotional reactions - likes, replies, reposts - Don't care about accuracy or seriousness because the algorithm doesn't measure it - Don't care about their existing audience because every post is judged in isolation anyway In short this new algorithm, like so many on social media, is all about maximizing whether people will engage with something - not about whether they should.

AI is making kids dumber. It should be making them geniuses. Introducing Koji, the first AI tutor that gets kids to actually think. 👇 Video

Windows users, this one’s for you. Computer use now works on Windows, so Codex can take action on your Windows computer. And with Windows support for Codex in the ChatGPT mobile app, you can start, review, and steer tasks on the go while work continues on your Windows machine. An early experience, but we’re working on more ways to keep your work moving, wherever you are. Video

Today, we're releasing LFM2.5-8B-A1B, a device-optimized model designed to power real-life applications on phones, laptops, PCs, robots, and fast & lightweight server-side use-cases. > 8B MoE, 1.5B active > Expanded 128K context > LFM2.5 flagship hybrid MoE architecture > Trained on 38T tokens + large-scale RL > fast, reliable tool calling, punching above its weight, comparable to models with up to 4x its size > customizable on a single GPU for any specialized task > LFM2 open-weight license 🧵

🚀PaddleOCR-VL 1.6 Officially Released! We are thrilled to announce the official release of PaddleOCR-VL 1.6 — this version has set a new SOTA record of 96.33% on OmniDocBench, outperforming both open-source and proprietary solutions in text, formula, and table recognition. 📊 Key Highlights: 🔸 Ranked #1 on OmniDocBench v1.5 and Real5-OmniDocBench as well 🔸 Significant improvements in table, classic text, and rare character recognition 🔸 Enhanced seal, spotting, and chart recognition 🔸 Fully compatible with v1.5 architecture — zero migration, plug-and-play From financial contracts and legal documents to research reports and historical archives — empower your document intelligence workflows by providing high-quality data to large language models (LLMs) and retrieval-augmented generation (RAG) systems. ✨ Industry-leading accuracy. Zero migration. Plug-and-play. 🔗 Read more: github.com/PaddlePaddle/Padd… #PaddlePaddle #PaddleOCR #AI #ComputerVision