stoikol

Support it if you want @Google and @antigravity to bring back Gemini 3.0 Flash. Keep 3.5, create 6.9 or whatever you want, but WE WANT FLASH 3.0 BACK

@leezardo2 @uzairakrum Updates and new models are perfect, and we all love them. But you shouldn't erase the basic models that are more cost-efficient. We don't need Lambos to go to the mailbox, a bicycle is enough. So, why you should eliminate them?

@air_codex @uzairakrum It's the same problem in computing in general, you don't get faster cpus or gpus, you just get bigger ones for a higher price, and obviously that means that the performance will be higher.

M3 is here on Opencode They have also updated their site And it looks like m3 will have signifcantly less requests compared to m2.7 So a big jump in api pricing

@MohamedHz72007 HER

@air_codex Been building an AI that "actually" has a life between our conversations. She simulates hobbies, a daily routine, messages me first sometimes. Been running for a day and already feels different.

if running a model this cheap can cost 2$ per day ( probably less when every thin is established and no heavy modifications) how other companies like google cover the usage of the 20$ per month pro plan with models much more expensive than deep seek and at that contact surface ?

@mikerubini nice, would be great scrape son descriptions, and create an index category about their niche

Every new website quietly announces itself the moment it goes live, weeks before anyone tweets about it. So I built a feed that catches products at the exact instant they attach a custom domain. New stores, new AI tools, new everything, in real time. Tool: intelligence.flatnine.co/launches Blog post: flatnine.co/blog/i-built-a…

Why are so many people starting to write everything full lowercase and removing "Capitalization" ??? I need to understand it. Can someone gimmi some context?

@gerard95033447 @NousResearch @stepfun how are you feeling the inference there sir??

刚在 @NousResearch hermes 使用了 @StepFun 最新开源的 Step 3.7 Flash，上手玩了几把，说下感受： 拿它当 Agent 用，让它自己去搜资料 → 读内容 → 写代码 → 出结果，一套下来流程没断。11B 活跃参数的 MoE 架构，响应速度确实快——那种"话说完就有回应"的感觉，用过卡模型的人都懂。 256K 上下文丢了一整个项目进去，让它做基于上下文的总结，该记住的都记住了，没丢东丢西。 接下来打算： 跑一个完整的自动化任务链，从需求拆解到执行全交给它自己串 试试 Agent 调工具的稳定性 看看有没有什么坑需要踩的 有也在折腾的朋友吗？你们测了些啥有意思的？评论区聊起来🤔 #Step3_7 #OpenSource #HermesAgent

@0xPocho congrats

Le gros travail en parallèle sur le SEO commence à porter ses fruits Checkez la landing page et les images d'articles de blog (je l'ai trouve trop belles), prompts générés via Claude et image via Gemini 👀 🔗 lien de la landing page en commentaire 👇

@amar_patel @theo @Teknium i think so yeah, money as interest layer is something quite relevant under "thermodynamics law"

@air_codex @theo @Teknium Yup! It is more accurate than the news sometimes!

wtf do you mean there's a pre-bundled Polymarket skill in Hermes Agent???

@supezen congrats !! 6 days is like a great achievement! you are fast

6 天时间 3.4 万行代码，从一个 idea 到第一版产品 NewFlow 内测版今天上线‼️

@amar_patel @theo @Teknium yes polymarket is not just for betting, they are crowd interest, that has value

@theo @Teknium You’re the absurd one today, Theo. Just disable the fucking skills - there’s nothing to whine about. Polymarket is a very common use case. You also baselessly accuse them of bad support. How would you appreciate that coming from another founder? I bet you wouldn’t.

@theo @Teknium we can delete the ones that are not in our development interest

@Teknium Bro there was a baked in polymarket skill you can’t possibly justify this as user error I don’t want an “empty soulless experience” I just didn’t expect a gambling skill to come along my super cool agent experience 🙃🙃 I heard yall are bad at feedback but Jfc this is absurd

@KSimback @kevincodex @gitlawb do you know the tokenomics? it's 100% over circulating

Earlier this week I spoke to @kevincodex about the @gitlawb project It was an interesting time bc the token had just dropped significantly but we mainly spoke about product I can say Gitlawb is one of the most ambitious projects in AI x crypto It’s basically a decentralized version of GitHub + Lovable + an open source Claude Code All 3 have some initial traction, and I’ll be interested to see how Gitlawb (the decentralized GitHub built for agents) does as staking and nodes go live Just sharing that after my interactions with Kevin I can say he appears to be a genuine builder in this space Disclaimer: I hold a small amount of the token because I want to support builders like him

@kevincodex Sir, I love what you are doing, but your UI needs serious work. All those fonts in your header are almost unreadable.

@corywilkerson @Cloudflare nice!!

This is awesome. Built a lil' extension that uses @Cloudflare Browser Run quick actions to extract markdown, links, html, take snapshots, screenshots, etc.

I'm experimenting with Hermes Desktop by @fathah_cr, working on two versions of the offices, one for 2D and another for ASCII with an integrated console. I hope to contribute these templates on GitHub soon. If anyone wants to play around, please let me know.

@hqmank nice, i was testing today, desktop and workstation, so now i have ti try portal as well. sir why are you using portal? just to get the combo all in in th same place?

Hermes Agent added stepfun/step-3.7-flash:free 🎁 I fed it screenshots, multimodal holds up. The model genuinely costs $0. Catch: you still have to add a credit card and "buy" the $0 subscription on Nous Portal. Free model, $0 plan, full card-on-file. Make it make sense.

im in

这几天在 X 上看到几个关于 /goal 的讨论，质量不错。 我挑了其中两个我认为特别有价值的模式，整理进了仓库，也在这里简单分享一下原理和实际价值。 原帖链接我会放在后面。

OpenAI 和 Anthropic 最近都在推 Workflow / Threads，但我觉得他们真正想解决的，是同一个更底层的问题： 当任务变得足够复杂时，单个 Code Agent 几乎必然会失败。 不是模型不够强，而是机制上就做不到自我纠正