FerTech 🇨🇭

@simonw Hey @Prince_Canuma I think here is where you guys come in and do your magic with MLX 😉

Dan says he's got Qwen 3.5 397B-A17B - a 209GB on disk MoE model - running on an M3 Mac at ~5.7 tokens per second using only 5.5 GB of active memory (!) by quantizing and then streaming weights from SSD (at ~17GB/s), since MoE models only use a small subset of their weights for each token

@elonmusk When are we going to be able to use it for coding included in our plans?

Try Grok 4.20 Heavy

@uiux_harshit How much time it took to each?

Google Stitch vs Claude vs Human Pick your winner

@TeksEdge @thrawling What “affordable” hardware would you buy today to be able to run effective models locally for vibe coding ?

@thrawling Was running 35B when posted. Besides the post wasn’t meant to be a performance controversy. Still in awe of LLMs. Rumors are BigAI will jack up prices by end of year and it got me thinking about local inferencing.

💡 It's still amazing to me that you can run an unsloth version of Qwen3.5-27B on a $2K AMD Ryzen Max+ 395 w/64GB of unified memory @ 10 tps at home. Nearly the same quality as Claude Opus 4 (May, 2025 release)

We just released Claude Code channels, which allows you to control your Claude Code session through select MCPs, starting with Telegram and Discord. Use this to message Claude Code directly from your phone.

@OfficialLoganK @GoogleAI @GoogleAIStudio @antigravity @Firebase @patloeber Also, what is the future then? Would antigravity come to the web and you guys are going to start unifying products? This would be awesome. I just vibe coded an agent with voice and retrieval via voice on my own data in 15 minutes this is crazy!

@FerTech @GoogleAI @GoogleAIStudio @antigravity @Firebase @patloeber correct, but we will start taking into account Pro and Ultra subs soon!

We’re launching a brand new, full-stack vibe coding experience in @GoogleAIStudio, made possible by integrations with the @Antigravity coding agent and @Firebase backends. This unlocks: — Full-stack multiplayer experiences: Create complex, multiplayer apps with fully-featured UIs and backends directly within AI Studio — Connection to real-world services: Build applications that connect to live data sources, databases, or payment processors and the Antigravity agent will securely store your API credentials for you — A smarter agent that works even when you don't: By maintaining a deeper understanding of your project structure and chat history, the agent can execute multi-step code edits from simpler prompts. It also remembers where you left off and completes your tasks while you’re away, so you can seamlessly resume your builds from anywhere — Configuration of database connections and authentication flows: Add Firebase integration to provision Cloud Firestore for databases and Firebase authentication for secure sign-in This demo displays what can be built in the new vibe coding experience in AI Studio. Geoseeker is a full-stack application that manages real-time multiplayer states, compass-based logic, and an external API integration with @GoogleMaps 🕹️

@OfficialLoganK @GoogleAI @GoogleAIStudio @antigravity @Firebase @patloeber That would be much appreciated. Also I reported a bug, I was all into the app and suddenly even with an API I got rate limited! Made no sense and I couldn’t continue working on the idea 😔

@poe_platform They are coming so fast that it’s difficult to keep up 😅

MiniMax-M2.7 is live on Poe! A next‑gen self‑evolving model built for autonomous software engineering and agent workflows. M2.7 can iteratively improve its own agent scaffolds, optimize task performance over repeated runs, and deliver major gains on real-world coding benchmarks like SWE-Pro. Great for multi-file codebase edits, long-horizon planning, tool use, document automation, and complex multi-agent tasks. You can try it in Poe app on all platforms and in the Poe API at poe.com/Minimax-M2.7

@elonmusk @diana_dukic Drastically improve support. Many of us got accounts suspended without any logic explanation. Some of us even paid for an advertisement got charged and then the publication was never approved by you kept our money, another nightmare to try to ask for a refund.

@diana_dukic What needs to be better?

Went from scrolling 24/7 to not even wanting to log in. X just hasn’t been hitting the same lately.

@UrbanaPlayFM Pero @bilinkis , probaste ChatGPT 5.4 Thinking???

"CHAT GPT ES EL NOKIA 1100". —Santiago Bilinkis

Oh yeah! @poe_platform now supports multiple API Key creation! Thank you guys! This was so much needed! Hope we see more enhancements on this direction like granularity of spend per API key so we know how much each project has consumed, etc. 😉 YOU GUYS ROCK!

So? What is going to be @Google´s next step? kill antigravity and integrate it in Google AI Studio? Keep them separate? Launch a new product? make your bets..

@elonmusk I got thousands of those puppies waiting for a nice valuation 😉

To be honest, this is typical Google, always doing things in a very confusing way, duplicating streams until their own internal battles show a winner...we as customers and users are caught in the middle... Imagine how much more efficiency they would bring to their products if they would all work together on the same project...

@FerTech @GoogleAI @GoogleAIStudio @antigravity @Firebase @patloeber Anti-gravity was locking pro users out for a week after 5 hours of use, just yesterday. Surprised they have the nerve to roll this out today while Pro users are still screaming about the pricing scam

One of the most useful services now has a full fledged app, congrats and thank you @Tailscale

@alesssj4 @GoogleAI @GoogleAIStudio @antigravity @Firebase Yup, already testing 😉

@GoogleAI @GoogleAIStudio @antigravity @Firebase @FerTech

So, @windsurf announced a change of plans, which come into effect TODAY! windsurf.com/blog/windsurf-… I asked ChatGPT if these are better or worse, here is the whole analysis so you dont have to do it yourself. There are still some gray areas I think, it would be good if the people from @windsurf confirm if the analysis is accurate or not. @WindsurfCurrent chatgpt.com/share/69bc232f…

Here is a quick walkthrough of everything new in Stitch: 🎨 The AI-native canvas can hold and reason across images, code, and text simultaneously. The new agent manager helps you design in parallel. (PS … light mode!) 🧠 A smarter design agent now understands your entire canvas context. You can swap images, generate product briefs, or mix mobile and desktop screens on the same canvas. 🎙️ You can vibe design with your voice (in Preview). Stitch can ‘see’ your canvas and your selected screens. You can ask for design critiques, variations, or navigate your canvas. ⚡️ Instant prototypes. Just hit the play button to see a prototype or preview your app in seconds. Stitch can imagine the next screen based on your mouse click. 📐 DESIGN.md and consistency. Every new design automatically starts with a cohesive design system which vastly improves consistency. The new DESIGN.md file can be used to export or import your design rules. Walkthrough video👇 Details for each major improvement 🧵

@baanditeagle @geminicli @rseroter Yeah, I am a paying user too and the bloody thing is too restrictive. Makes no sense, I am constantly being pushed to use an API. I am already paying something, that should be worth more than just a couple of times of usage.

An ultra user cannot even build something in one go without waiting for hours of limit! Forget about these free tier and pro accounts! You guys are giving everything everywhere and it is making us not use one platform completely. Instead of giving different limits on different products, isn’t it better to have one single limit on all tools? Example - gemini cli and antigravity? I started working on an extension and by the time I could half test it, I was out of my limits on an Ai Ultra account.

We want to be transparent that we’re making changes to Gemini CLI that may impact your workflow, including more robust detection of policy-violating use cases, prioritizing traffic for certain types of accounts, and restricting Gemini Pro models for free tier users. We understand that these changes represent a significant shift, and apologize for any disruption this might cause you. For direct control over your own quotas and billing, use your own paid API key from AI Studio or Vertex AI. We’re here to support you, and will do our best to address your questions and concerns. As always, thanks for building with us. 🙏 You can read more here: github.com/google-gemini/…