Charles Cai - e/acc

@sch which browser-use plug-in or skill you are using? thx

codex with a browser is wild. in 2m it found an issue in my Ubiquiti network config that was causing really slow transfers from my camera storage. it literally looked up the network topology, checked the switch & AP settings, and found the routing issue…and is now fixing it 🤯

@arkham @grok fact check, and what’s the latest news between WLFI and Justin Sun?

The WLFI Team is borrowing $150M USDC against $400M WLFI on Dolomite. The WLFI Team is lending $406.23M of WLFI across 2 wallets. That is 4.99% of the supply, and 97.8% of the WLFI cap on Dolomite. They are borrowing a total of $150M USDC against their holdings on Dolomite.

@Ali_TongyiLab Open weights releasing soon?

1/10 🚀 Qwen3.5-Omni is here! Scaling up to a native omni-modal AGI. Meet the next generation of Qwen, designed for native text, image, audio, and video understanding, with major advances in both intelligence and real-time interaction. A standout feature: Audio-Visual Vibe Coding: Describe your vision to the camera, and Qwen3.5-Omni instantly builds a functional website or game for you. Highlights: Script-Level Captioning: Generate detailed video scripts with timestamps, scene cuts & speaker mapping. SOTA Performance: Qwen3.5-Omni has secured 215 SOTA scores across various sub-tasks, matching the top-tier text/vision capabilities of the Qwen3.5 series. Audio-Visual Understanding: From auto-segmentation to fine-grained script generation, it understands the relationship between characters and their environment like never before. Seamless Interaction: With native API support for Semantic Interruption, voice conversations feel human-like and background-noise resistant. Global Multilingual Mastery: Pioneering support for 74 languages in speech recognition and 29 languages in expressive speech generation, breaking down global communication barriers. Autonomous Intelligence: Native support for WebSearch and complex Function Calling—the model now independently decides when to pull real-time data. Qwen3.5-Omni is built to be the backbone of next-gen AI applications, empowering developers and users alike with true multimodal reasoning.

Limited beta testing, DM @qianjiang___ pls

@antirez Well, interesting, I thought it was the other way around... techcrunch.com/2026/02/23/ant…

May users reported that asking Sonnet 4.6, via the Anthropic API, "What's your name", it reports "DeepSeek" with a high frequency. Labs cross-fine-tune on chain of thoughts and use other models as RL signal consistently. Also the pretraining, a *key* step, is on mostly public data. Anthropic disappoints me deeply.

@wangray Qmdvsearch自带有reranker吗？能换embeding models吗？qwen3 两个都有，各是8B的 大是大了点肯定更准确

@V7700SN @grok English please

Les Samsung tu les ouvres un peu brutalement ils te lâchent mdrrr

@XFreeze @grok fact check

President Trump at the UN: "When your prisons are filled with so-called asylum seekers who repaid kindness with crime, it's time to end the failed experiment of open borders" "According to the Council of Europe in 2024, almost 50% of inmates in German prisons were foreign nationals or migrants." "In Austria the number was 53% of the people in prisons." "In Greece the number was 54%." "In Switzerland, beautiful Switzerland, 72% of the people in prisons are from outside of Switzerland."

@NVIDIAGeForce GeForce Day

🟢 GEFORCE DAY IS BACK 🟢 To celebrate, we're giving away TWO GeForce RTX 5080 Founders Edition GPUs, signed by NVIDIA CEO Jensen Huang. Want one? Comment "GeForce Day" for a chance to WIN & stay tuned for more!

@georgesttock NANO

Nano banana + MakeUGC + Veo3 = Ad Factory This agent creates 100s of ads every day - No actors - No product - No $10k/month agencies You're able to clone your comepitors ads with AI Paste there ad -> Pick an avatar and regenerate. Comment "NANO" and I'll send you the agent + the full playbook (must be following)

@MaximeRivest How does FP8 model work with 3090, which doesn’t have native FP8 support…

Did you know that with uvx & vllm you can run 1 command in your terminal and have the new Qwen 3 running on your machine? uvx --python 3.12 vllm serve \ Qwen/Qwen3-30B-A3B-Instruct-2507-FP8 \ --tensor-parallel-size 2 \ --max-model-len 8192 \ --gpu-memory-utilization 0.94 \ --swap-space 8 \ --enforce-eager And if you start OpenWebUI, you get to chat locally with Qwen 3 In this video I host Qwen/Qwen3-30B-A3B-Instruct-2507-FP8 ! This one is much less quantized then the other one in the post I quote below, so I need both of my 3090 gpus and it runs a little bit slower. HOWEVER go check the resulting website (toward the last 15 secs of the video). Qwen one-shotted and it is very 'impressive'. The video is real-time, you might want to skip the middle ;)

@ju3t1ng @Gorbagana_chain @lex_node Great - where’s open pose for squad? Can the app propose moves (showing recorded pose, display and gauge the user’s openpose, to detect: 1 - number of actions / speed of actions 2 - accuracy of the moves 3 - completeness of the moves (ie full height move not just half) …

#GorbaganaTestnet @Gorbagana_chain @sarv_shaktimaan @lex_node putting #fitclash on solana part 4

@ReallyAmerican1 @ProjectLincoln @grok fact check

BREAKING: In an insane moment, Donald Trump claims that Vladimir Putin lost 51 million people in WWII "and he did fight," delusional remarks to a room stunned into complete silence by his cognitive collapse. Make sure everyone sees his shocking delusion!

@grok please advice: Q1: What does short url duplication detection mean here - for those users specify a short name to use? Q2: Why the designer / architect thinks 10ms query on this would be a bottle neck if high concurrency of Redis is good enough? Please show some Redis performance metrics to verify this argument?

Designing a URL shortener for 100,000 unique URLs per second requires a scalable architecture. Key components include an API gateway to route requests, a URL shortening service using base62 encoding for unique codes, and a distributed database like Cassandra for storing mappings. To handle the scale, implement a caching layer with Redis to reduce database load, targeting a 90% cache hit ratio. Use load balancers to distribute traffic across 4-6 API servers and 10-20 database nodes. This ensures low latency and multi-tenant isolation, while monitoring systems maintain reliability. The diagram you shared captures the basic flow but lacks caching and load balancing, which are critical for high throughput and performance at this scale.

Designing a URL shortener sounds trivial until you're tasked with generating 100K unique URLs per second, with low latency, strict correctness, and multi-tenant isolation. Over the weekend, I explored a brilliant breakdown of how Rebrandly engineered for this scale 🧵

Thanks to Minghan Xiao from Imperial, whose team organized the hackathon, and @flock_io for the long-term support ahead! We really enjoyed the hackathon and are busy productionizing MeSu AI

@godofprompt @hasantoxr GPT

I turned ChatGPT into my personal assistant. Now, I work just 60 minutes a day, thanks to my GPTs. So, if you want to create GPTs in ChatGPT to automate all your tasks then I have created "GPT Creator Mega Prompt" for you. Like + comment "GPT" and I'll DM you the file. (Must be following to receive it)

#AI #STAC comes to London! My panel discussion is on "AI development stacks 🛠️", Leaner AI: Model to Metal; there are also talks on "Enterprise AI stacks 🏢", "AI training stacks 🧠", "AI inference stacks ⚡" etc! Registration is here: stacresearch.com/aistacLON #AISTAC

@MooenyChu @op7418 ChatGPT4 / DallE : second one enhanced with sunshine

@op7418 DALLE3出的，差距有点大

🧪Midjourney V6 每天都能给我新惊喜，一个玉石雕刻的国风场景，由于细节很丰富，看起来真的像工艺品。 昨天在小红书看到一个玉雕的图，但是作者没写提示词，今天复刻了一下，效果还好了很多。 主要是 V6 对提示词的理解太好了，之前一些可有可无的词真的不太影响效果画面描述准确就行。 提示词： miniature landscape, jade sculpture, mountains, a chinese ancient architecture are in between a cloud, in the style of gold and jade, photorealistic details,Very good light transmittance, miniature sculptures, made of jade, gold and emerald, uhd image, Epic, octane render, beautifully detailed, light diffusion, cinematic shading, cinematic elements --ar 3:4 --v 6.0

💚 #BRC20 + $Wom = 50% airdrop… twitter.com/wom_eth/status…