Introducing Mobbin CLI An unofficial Mobbin CLI for searching UI screens, flows, and sections right from your terminal. - Search screens, flows, and page sections from terminal - Built-in agent skill for claude/codex/etc - Download images locally for references

@theo @r_marked @t3dotchat I was using t3 chat, and it just keeps generating logos non-stop. IM ONLY BEING CHARGED FOR ONE.... THIS IS COSTING YOU GUYS TONS OF MONEY!!! Please repost and @ theo so he can fix this ASAP

The fact that I can't connect a existing github repo, and build from there

It'll just cost them money, I watch Theo's videos and run one of my own ai platforms. Models are expensive, this was the right thing to do!

Claude Code will be gone if Codex CLI will support your ChatGPT account and use that usage not just make an api key...

Gemini 3 will crush every other model, Gemini 2.5 has a problem with tool calls. But even after a couple months it's still #2 on LM Arena. The Kimi K2 release is probably going to give Gemini 3 good tool calls as well. Gemini 3 is gonna be fire 🔥

Glad to help Theo! :)

Nice

Once this comes out, vscode is being uninstalled

Switched to OpenCode form @thdxr, my reasons: - Claude code took over 8 seconds to start most of the time - Clunky ui when running multiple agents - Scrolling was painful But OpenCode looks way better, is way faster, and feels way nicer to use!

*Cough cough* Llama 4...

Yea sure: I need a logo icon for 'projly.app' that immediately communicates "seamless project management, progression, and uninterrupted workflow." This icon must be incredibly modern, truly minimalist, and very approachable. It needs to convey organization, efficiency, collaboration, and a friendly digital sophistication. **The Icon: A Softly Rounded, Luminous 'Continuous Path' Shape** The central element is a single, highly stylized abstract shape, conceptually representing a smooth, unbroken pathway or flow. This shape is crucial: it must feature **soft, subtly rounded corners and edges** throughout its entire form, avoiding any harshness or sharpness. It needs to look fluid, inviting, and inherently continuous. It should be rendered in a **perfectly flat, two-dimensional style**, with **no chunky, beveled, extruded, or faux-3D effects** whatsoever. The goal is a clean, smooth, and modern design. * **Form and Geometry:** The icon's silhouette should be a clear, single, thick, and gracefully flowing line that forms an abstract, continuous loop or pathway. Imagine a single, unbroken stroke that curls, perhaps crosses over itself once or twice in a soft, elegant manner, and appears to lead onward. Its lines should feel fluid and organic, not rigid or fragmented. The entire path should have uniform thickness with perfectly rounded end caps (if it's not a closed loop) and smooth, pleasingly rounded transitions at all bends. It should convey a sense of gentle forward momentum and cyclical completion. * **Luminosity and Color (Crucial):** The entire solid shape of the path must be filled with a **single, soft, and luminous green**, employing a **subtle, harmonious radial or multi-point linear gradient** that appears to **radiate light from within its own form**. This is paramount: the color should appear to **glow internally**, creating a soft, inviting radiance. It should not be a harsh or jarring multicolor. The green should transition from a deeper, rich emerald or forest green at its base to a brighter, soft lime or spring green towards the areas intended to feel most illuminated, ensuring the glow feels integrated and not a separate effect. The goal is a unified, soothing, yet energetic glow, evoking growth, clarity, and continuous progress. * **Absence of Outline/Border (Implicit Sharpness):** There must be **no separate, visible outline or border** around the icon. Its defined form and soft contours should come purely from the crisp edges of its luminous, solid fill. The transition from the icon to the background should be seamless and clean. **Overall Aesthetic and Mood:** The final icon must seamlessly integrate with your UI's **simple, minimalistic, and nice aesthetic**. It needs to look high-quality, friendly, and inviting, while still conveying organization and sophisticated project management capabilities. It should feel approachable and trustworthy. It's essential for this icon to be highly scalable, looking clear and impactful as a tiny favicon (where its soft, luminous shape and unified color will be very clear) and equally impressive as a larger graphic on your website. **Crucial Elements to Strictly Avoid (to prevent past issues and ensure desired result):** * **No Hard, Sharp Angles or Jagged Edges:** All corners and lines must be soft and rounded. * **No 3D/Bevel/Extrusion Effects:** The icon must be entirely flat and vectorial in its appearance. * **No Chunkiness or Heaviness:** The design needs to be sleek and light. * **No Generic Clip-Art Look:** The icon must convey custom refinement and a friendly elegance. * **No Fuzzy Blurs or Diffuse Glows (unless soft and uniform, as part of the internal luminosity):** Any luminescence should feel integral to the shape, not a separate effect. * **No Outlines or Strokes:** The shape's definition comes from its solid, luminous fill. * **No Multiple Disconnected Elements:** The entire icon must be a single, unified, flowing shape. * **No Backgrounds or Additional Elements:** The icon should stand alone on a black void to maximize its luminous effect. * **No Text or Letters within the Icon:** The icon is a purely symbolic graphic. * **No Jarring Multicolor Gradients:** Stick to a single, harmonious color or a very subtle gradient within the green spectrum. * **No Literal Project Management Imagery:** Avoid literal gears, cogs, stacks of papers, or overly complex flowcharts. The icon should be abstract and conceptual.

Going to make a new project, and I'm trying orpc! Really hope this is the right decision over trpc 😅 Using @amanvarshney01's better t stack for this!

Better T Stack = My Stack Also switching the following: Biome --> Oxlint (Way faster, and less resource usage from my testing) Trpc --> Orpc (Faster and more features)

Claude Code is down right now! Seems to be constantly stopping and not responding to messages. Hopefully they fix this soon!

So clearly my maintence pages get tons of traffic since I have many users, sorry guys I'm working on getting everything up! But the crazy part is someone tried to ddos me, but @vercel firewall easily defended it. Thank you @rauchg ❤️

I JUST GOT ACCESS TO PERPLEXITY COMET!!! 🔥 If I get invite codes I'll be dropping them!

who's gonna tell him that @Cloudflare lets you do this for free 😭

Why pay money for Grok 4? Just use it for free on gl1.chat

Who do I believe??

Claude Opus 4.1 spotted! Hopefully they also drop a Sonnet and Hiaku version. I think it'll come in 1-2 weeks

After a few days of using Codex... I don't recommend switching, the experience is a lot worse than Claude Code's. I already got a refund and switched back to Claude. I feel like this update was overhyped and Claude Sonnet 4 is still the king of coding!

Ok wtf is going on with @cursor_ai I was just using o3 and now I can't use it! I still have my subscription this is so stupid. The past few updates ruined cursor, Claude 4 Sonnet rate limits are even worse then 2 days ago....

This is so real, the first day I had unlimited usage of Claude 4 the next day it was terrible and it's just terrible now...

@openchatpro🤝@convex OpenChat will be built with Convex. We will not be using D1 DB as planned, here are the reasons: - D1 is hard to make real time - D1 will be somewhat expensive to scale - D1 is hard to scale Looking forward to this switch, we will still support sqlite if you self host (will try to make this work)

bunx create-better-t-stack, is way better

Some more details: I was first going to use @convex then I found out about eletric sql, but in the end @convex is my choice My reasons: - Electric SQL is hard to set up - Convex is super easy to use So @openchatpro will be run on @convex Great job Convex Team! 🔥

Lot's of work these past 2 days for @openchatpro Some things I added: - OpenRouter Integration (Login with your account and use any model from there) - Smooth streaming, stop response, resume response - Nicer ui (looks nicer!) Going to be taking today off, I'm tired 🥱

Added a sidebar to @openchatpro Also working on the ai integration Demo of the ai integration soon 🔥

Going to be dropping a cool tool tomorrow that I built today. Imagine fumadocs but for roadmaps! You can quickly make a roadmap (takes 5 minutes) It uses ShadCN (bring your own theme) First looks (default theme):

@v0 @rauchg we need this please 🙏 This is the only thing stopping me and many others from actually using this in places that are needed

Reached 100 followers ❤️

When is @AnthropicAI going to fix Claude Code, like it's the best cli tool but weird problems always happen... Am I the only one or is this for everyone?

From a mockup to a product W design?

2 weeks of work pays off! My account has been growing a lot for the past few days, thank you to everyone ❤️

Not there yet only at 30 lol but I post consistently... Idk I feel like at 1 point i'll go viral :)

Before I'm too deep into building @openchatpro I'll be switching my stack, to this: - Next.js - Elysia - oRPC - Postgres - Drizzle Some core features I will add with this: - Local usage - Realtime sync - Self hosting I will release a non beta version before December 31st!

Just gonna say it: Claude Code gives more value than Cursor at the this point

Claude Code + @bunjavascript = ❤️

I have come to the conclusion that releasing gl1(.)chat won't work :( My main problem is: Big companies like OpenAI have pretty much unlimited money and can offer way higher limits. I will probably continue the project and make it open source! This is the sad truth but it's the only possible way to still continue my passions. I will be open sourcing it in the next couple of days, will post the github link and the official website soon! Any contributions will be welcome :)

Is there really no fast browser written in rust or smth? (i fr might build my own 😭)

My account for the past 7 days! 📊 26K impressions (+139%) ⚡ 443 total engagements (+45%) 📈 1.7% engagement rate

I have the Claude Code max plan, but I hit the limit yesterday 🥲 (My usage for just yesterday below 👇)

Update on this: OpenChat will be run on Railway with @ElectricSQL Been on railway a couple of days and it's looking pretty cheap... But I spun up a electric sql instance (only that) and it's estimated to cost 3$ Starting to make me thing that it'll cost more

When are we going to get an ai model that actually disagrees and doesn't sugarcoat? Like hearing "Oh that's a great idea" or something else when it's clearly a terrible idea is annoying. I would use a model like this every day

Just gonna say it: Github copilot is actually way better than Cursor for the price. While Cursor gives you 20$ in credits to spend, github copilot gives 300 messages which can be very long conversations with Claude 4 giving you way over 20$. It's also 10$ cheaper 🥲

Going to also try to host on @Railway, later might switch to VPS if it's to expensive (apparently people save tons of money) @convex also has a easy deployment with railway so that's really nice!

I'm still waiting for Grok 4 code :p

gl1.chat that's the real answer

Redesigned the UI a bit! Be honest is this a good ui?

Only like 10,000 seconds = around 3 hours on Claude code 😅

We got a nice ui now, building out the functionality next, later the cloud features (cloud syncing aswell as ai features)

gl1.chat launched 4 days ago and we already have: - 6 users! - 24 chats made - 64 messages sent - 7 images generated Not bad... cool thing is that gl1.chat is free! Try it today and tell me what you think

My ide journey 😅 VSCode --> VSCode with Copilot --> Cursor --> Trae AI --> Zed --> Kiro --> VSCode with Copilot Honestly I'm happy that Copilot is way better than before!

gonna use better t stack to rewrite projly.app also gl1.chat is powered by better t stack. thank you so much @amanvarshney01 ❤️

Just gonna say it: o3-pro isn’t gonna be as exciting now. Gemini 2.5 Pro is beating o3 at such a low price! Even if o3-pro is better, it’ll probably cost way more—just like what happened with o1-pro. #AI

I want a fast browser, not like chrome and stuff, I want zed.dev but as a browser fully rust based and crazy fast.

This is why I use better t stack ❤️ Also at this point with AI Agents you can easily add the stuff @shipfa_st has

Go burn those tokens! 🔥 Benchmarks on SWE-bench: - Claude 4.1 Opus: ~74.5% - Claude 4 Sonnet: ~70.4% - GLM-4.5: 64.2% - GLM-4.5-Air: 57.6% Not a full replacement but close to it, maybe good for running those 24/7 loops or small tasks

@rauchg any way to host on a vps or something else and use vercel with bot protection?

T3 Chat vs Perplexity Image gen (GPT Image gen selected for both) Clearly @theo offers high or medium quality while perplexity (20$ a month) gives low quality

Claude please go add this component and put it on this page I get this back... (Not the component, some weird ai generated crap...)

I'm having some deployment problems, I tried my best today to fix them but it might take a bit 🥲

Flow from DriftJS isn't that cool yet. But the 1,266 using it get safer and more performant migrations. driftjs.com/flow

What are unique products that I should use for @openchatpro (Anything that will help build a open source ai chat) Already will be using: - Formidable from @DimitriGilbert - Gradient generator from @Inveeest

Btw I'm building this for 2 core reasons: 1. I want to write down my ideas and todos somewhere and not have to load a heavy website each time 2. My computer was shut down while I was writing a prompt into Claude Code and I can't bring it back so I want something that automatically saves on each keypress with no delay

It's crazy that ChatGPT is winning by alot... Like their website is pretty slow and clunky. I would personally choose Grok out of these options, or if you want to go for an option that's not on the list you could choose @openchatpro The ui is pretty good:

Fast Good at code (from what I'm seeing from feedback) It's gotta be gemini 3 🔥 (My guess seems to be correct for now!)

I've been shipping and Someone is only sleeping

Got @convex working on @openchatpro Craziest part: Took me 2 minutes 😭 Postgres usually took some time to set up, this was so simple. If you haven't tried @convex. You are missing out

gl1(.)chat will be released soon! It's not everything I Promised and I'm sorry but I've had some errors and I just want to get the main product out! I'll be dropping a 40% pro discount soon to let everyone try it. Thanks for the support ❤️

v0 agent is very cool! I tried it out and it's way better at making ui's now, I even decided I want to use it in my projects. I went to connect my github account but you have to create a repo no way to connect an existing repo and use v0 there? @rauchg this would be a great feature and instead of making starting ui's v0 could actually help build real projects!

GPT 5 released 26 days ago... @Trae_ai still doesn't have it Pricing also doesn't make sense, Deepseek V3 being cheaper than 2.5 Flash is a premium model... and r1 isn't that expensive either

Most annoying thing with Claude models?

5 more days... Use Claude Code a ton before the rate limits hit Hopefully they won't be totally bad and what they say is actually true

I've made it guys ❤️

Grok 4 is coming out today! Would be cool to have something like a Claude Code but for Grok 4. Has to have very high rate limits though or I'm not switching lol

Not sure if anyone noticed but it seems like Claude 4 Sonnet isn't filling it's context length up as fast on @cursor_ai I suspect its because now the codebase searches don't search entire thing and they have another model do it for the model and give the required lines of file.

GPT‑5 for coding but here is the pro and cons: Pros: - Plans well - Follows instructions - Doesn't hallucinate at all Cons: - Slow (compared to Sonnet 4) - Burns more tokens - Gets stuck in loops still I still use Sonnet 4, what's your experience with GPT 5 so far?

hmm instresting... 👀

Just got a mac and wanna try new apps! Drop your app link below and I'll try it (maybe even buy your pro plan if its useful to me) 👇

The fact that there is upload limits is crazy... On gl1(.)chat I am going to offer a simple pricing plan: 5$ = 10 million tokens across many models You can upload files, search web, do anything you want with your tokens. If you run out you can buy some more or upgrade to 20$

*gulp* might have to buy more premium credits on t3 chat 🙃

gl1.chat has real-time sync now! It's still in early beta and not garunteed to work, if you have any problems or questions dm me :)

Took me like 5 minutes to make a nice input for OpenChat with @v0 - Nice animations (not stupid ones) - Modern ui - Not 10/10 designer level but some manual changes and this will be great Overall a nice 8/10 already! Any suggestions on what to add to this?

I'm adding real time sync to gl1(.)chat Still hasn't launched yet, fixing some prod errors and making a launch video. Not sure when I'll launch it but it'll be a perfect experience when it does! Any suggestions on final features to add?

finally fixed it you get 5 free image generations per day with the best imagen model from google on gl1.chat!!!

Claude 4 only got 2 weeks and google beat them already

I have the feeling of a full on agent which is similiar to Convex's Chef since making nice components limits your market reach?

i've been using this and it's crazy how good the results are... this is like 5 times more effective than just using cursor alone...

Hopefully Llama 5 will be better

this is kinda late but gl1.chat got added to better t stack showcase!! thank you @amanvarshney01 ❤️ link: better-t-stack.dev/showcase

Just use better t stack and host on cloudflare/vercel/railway

First post to get over 1k views! It's 8k views and I'm so happy :)

Hi Opus 4.1 It's on the website, probably on the api or will be added to the api!

Only 9 more followers until 50 followers!

nevermind they finally responded and unsuspended me, thanks @useblacksmith

ok now thats really cool, i was thinking of something similar to this for gl1.chat

Is this the Sonic model? This was so unexpected though, I will be trying it and giving my opinion soon! Tell me what you think of it

For anyone wondering, it's a demo project I wanted to try to make with orpc. It's also a feature I might wanna add to gl1(.)chat! It's kinda like lovable but it is specifically made to make very high quality premium designs!