Rexx DZN

@LyalinDotCom Lol this is brutal. Even if u get the latest mac or that other thousands of dollars machine u get a sub 10 t/s generation speed. Local llm is not yet optimized enough to be replacing the cloud

@Movchanets @malikwas1f Dont get this

@malikwas1f credits not tokens

Holy! Those are some serious plans from Mimo right now.

@artpi @thekitze For how much

@thekitze They DO have an unlimited plan! I’ve been using it past month at work and they even hide the usage. chatgpt.com/codex/enterpri…

5th one... i really really dont think its that hard for openai to be like "hey man here's infinite credits you have been yapping about us for a year, also here's a fucking basketball"

@cn_siva @dealignai Woah. Is that the fastest dense u tried? Or do u use mor

@RexxDzn @dealignai with M1 Max 64 GB for the Qwen3.6-27B-MXFP4-CRACK-MTP

Qwen3.6-27b and 35b MXFP4 MXFP8 CRACK is out now with MTP. Enjoy uncensored speediness! 35b mxfp4: huggingface.co/dealignai/Qwen… 35b mxfp8: huggingface.co/dealignai/Qwen… 27b mxfp4: huggingface.co/dealignai/Qwen… 27b mxfp8: huggingface.co/dealignai/Qwen…

@TheAhmadOsman One thing that isnt talked about is the electricity cost. Isnt it like buying an rtx 3090 every year? So then whats the point in running locally if cloud is continuously becoming bigger, better, faster? Seems like a bad investment like the bought m1 64gb ram i bought bcs of my TL

You should buy an RTX 3090 and learn how to run models locally The elite don’t want you to know this but running local models is hella easy, performant, and cheap nowadays

@thekitze Design wise looks technically better standalone but in a sum those are app icons not icons for menu ... Looks even more sloppy than before lmao

ok maybe i should kill your hype a little bit

@0xMaxSF @0xSero 100 percent agreed. The costs carry it way too much

@0xSero its pretty bad. clearly benchmaxxed - doesn't have focus or taste for most tasks. Mogged by composer. For self hosted stuff i somehow find that GPT-OSS-120b is the most performant and also the most sane, even compared to new dense qwen/gemma models

What’s your experience with deepseek like?

@TeksEdge @lmstudio 10 tps is NOT usuable unless u are only doing very small work that u can run overnight.

💡 Cheap Localmaxxing: Currently running Qwen3.5-4B MTP Unsloth Q4_K_XL (reasoning) on a 3-year-old 👴 Ryzen 5 PC with iGPU @ 10-15 tps w/tools & cx 64K. Very usable speed even with tools (Web Search, Scraping) ⚡️ @lmstudio w/MTP 💪 🏆 > GPT-5.4-nano, GPT-5, o1-pro

@chayprabs @thegenioo @cursor_ai After 80m it used 1 percent?

@thegenioo @cursor_ai Lmao, it didn't moved till I used like 80 million tokens.

Is this bugged, am I tripping or what Because I am using Composer 2.5 decently now but no effect here so far

@Kappaemme1926 No i can confirm as well... Actually feels like 68 percent faster drain

I’m convinced Codex limits are draining way faster than before. Am I tripping, or did something change?

@thekitze @supermac_app Lol how many codex accounts do u have. Make it replace ticktick. jk bet thats not possible with all the features in it.

i hate that there's no calendar quick glance in the menu bar so i added it to @supermac_app. one less external app to install. what else u want

@jxnlco Android and windows Arm 64 native support

If you're using codex desktop app today, what features do you feel like are still missing? Let me know and I’ll summarize all the feedback and share internally.

@thekitze @supermac_app Nah no way ur doubling down. First of on mac there is some recurrent color for u its all random. On M there is. Thicker icons, readable, mostly filled not mere outline. Ur colors are overly saturated reminds of slop and the gradients chosen.. i can go on and on

@RexxDzn @supermac_app ok

can't believe i was hating on liquid glass when it launched i can stare at this all day 😍

@jun_song This is really sad

To my 12,000 followers, I completely understand if you are confused or disappointed by my sudden posts about crypto. Those who already have strong negative feelings probably won't accept anything I say right now. However, I have spent the last few months deeply studying the open source ecosystem, meeting countless developers, and listening to their struggles. The reality is that the open source funding system is completely broken. The top 0.1% of high-profile creators hoard the vast majority of sponsorships, regardless of how critical their project actually is. The rest of the developers are barely getting by, even though many of them have made breakthroughs that could literally change the world. I was incredibly frustrated by this broken system, and I kept searching for a way to fix it. I came to the conclusion that crypto mechanics are actually the most effective tool for this. So, I decided to risk my entire reputation and face all the stigma head-on to flip this into a positive ecosystem. Just like I previously worked to change the narrative around local LLMs and Sovereign AI when they were dismissed. Open source must win, and we don't have much time. If you are too disappointed, you are welcome to unfollow me. But I will welcome you back later, after I have radically revolutionized this deeply rooted problem. I am sorry to have let you down. Sincerely, Jun Song

@TechieUltimatum But that means let's say my S26u won't be able to connect with tab S7+ since the tab is on oneui 5.1 or so

Google is building Android’s answer to Apple Handoff 🔥 Android 17’s new “Continue On” feature lets you seamlessly move tasks between devices on the same Google account. What it does: • Start on your phone → continue on your tablet • Resume apps instantly from the taskbar • Works across Android devices • Built directly into Android 17 Google is finally turning Android into a true connected ecosystem. 👀 First testing arrives with Android 17 RC1.

@jeetnirnejak @paper @motiondotdev @zeddotdev @runfusion inspo

Sprint planner agent: don’t create/move your tickets manually, let your agent take care of them Designed in @paper, animation with @motiondotdev inside @zeddotdev

@cubic_wu @nicekate8888 @antirez Could be used on 64gb no?

@nicekate8888 @antirez 妈蛋，心慈手软买了64gb的

我的 Mac 已经跑上 DeepSeek V4 Flash，感谢 @antirez DwarfStar 4 + ds4-agent / OpenCode / Open WebUI 体验都不错 设备：M4 Max 128GB 模型：q2-imatrix 本地实测：ctx≈27.7K，gen=3.95K，generation avg 24.22 t/s ds4-agent 亮点：低延迟、KV cache 状态更稳，而且工具调用和 system prompt 都是围绕 DeepSeek V4 Flash 垂直设计的

@0xSero 93 is actually not that much for ur size

For your own mental health I’d suggest muting words and people. If people are particularly egregious, block them.