Aaron Li @polymorpher

Steam Frame is here! Our wireless VR headset + controllers, available now in the following models: Steam Frame 256GB Steam Frame 1TB Sign up now and learn more on @Steam store.steampowered.com/hardware/steam…

Tried polymarket.us at Blizzcon (Overwatch Championship) for the first time - fast, smooth, crypto invisible throughout the whole user experience

@sjdedic It would be helpful to have a link to the case, so we can understand the context and what exactly the founder was allegedly doing. Arrest, charge, court case records are public so there really isn’t much privacy concern.

@cursor_ai @bot Much needed - would still be nice to manually manage some agents and give them limited context (traditionally achieved by tagging) for some review and audit tasks

Is anyone going to create a vision pro successor? Rumor says the project was cut, teams moved away. I have been a daily active user since its launch - albeit mostly for the overlooked use cases: cardio, personal training (zoom), private display in public space

Gemini 3.8 Flash is now available in Cursor!

@jiahanjimliu why not just use an API key? seems just some extra cost issue that will happen in November, unclear if quality is going to be affected in long term

@arbiterstudio @Palworld_EN Bought the full set, would be great to have more cattiva and depresso options

Sharing confidential information with AIs is a norm these days! We can do better. Our latest encoder models find and strip 40 types of PII across 16 languages, all in one simultaneous pass for a secure information sharing with AIs. LFM2.5-Encoders identify all entities in the text at once. You can choose the entities yourself. They run inside your pipeline, so your private data never leaves your environment. No generation, nothing to parse.

@HTX_Studio Much better than Google's nap pods (retired)

What if we can print a fixed-weight board at home one day cheaply, and select arbitrary models just like what we do with 3d print and filaments

Sasha Malysheva @aimalysheva

a month ago

on Thursday, @taalas_inc agreed to join AMD they etch a model's weights directly into the transistors, so there's no external memory to read during a forward pass and the memory wall that caps every GPU inference system doesn't apply: their demo chip runs Llama 3.1 8B at ~17k

1 1 24 9K 9

Today, we’re excited to open-source TwiL-LM3, the first formal reasoning model from the webAI Intelligence Lab. At just 3 billion parameters, TwiL-LM3 outperforms OpenAI’s GPT-OSS-120B on 4 of 5 formal reasoning benchmarks while running efficiently on consumer hardware. That’s 40× fewer parameters, 2.6× faster inference, and state-of-the-art performance in the reasoning tasks that power reliable tool calling, code generation, structured outputs, and AI agents. TwiL-LM3 was trained using webAI’s proprietary reasoning pipeline on webAI-owned, verified datasets—not scraped internet data. We believe better reasoning comes from better training pipelines and higher-quality data, not simply larger models. Our approach demonstrates that efficient models can rival—and in many cases surpass—models dozens of times their size. Designed for the edge, TwiL-LM3 runs on hardware people already own—from a Raspberry Pi to an iPhone—bringing advanced reasoning to millions of devices without relying on the cloud. This is our first open-source release from the webAI Intelligence Lab, and it’s only the beginning. Proudly built in Austin, Texas. Article: webai.com/blog/webai-rel…

Today we are introducing Dyna-2, a world-action model pre-trained on one million hours of human video. At this scale, for the first time, we discovered several new scaling laws: • world-action models exhibit scaling law on human data across four orders of magnitude, from 1000 to 1,000,000 hours, • this human data scaling law implied a scaling law on never seen robot data, • both data and objective matter; world modeling and scaling on video data are essential for cross-embodiment scaling transfer to emerge 🧵

🚨 GAME CHANGER In April, we launched a NEW CAMPAIGN to recruit video gamers as air traffic controllers — and supercharged the entire hiring process to get the BEST & BRIGHTEST in faster. The results are HISTORIC ✈️ ✅ 94% of hiring goal met — fastest time ever to reach target ✅ 2,000+ hired — the most in a single year ✅ 2,000+ “best & brightest” candidates in the pipeline — reaching the Academy quicker than any class before them, thanks to the most streamlined hiring process We’re building the strongest, sharpest workforce in aviation history ✈️

DEFCON: New Red Team Tactic doctoreww.github.io/EvilFontTool/

@hosseeb @nucleusgenomics I had LLM writing code to test for short sleep genes BHLHE41/DEC2, ADRB1, NPSR1, GRM1 - also find genes from latest research papers on specific areas (like muscle growth and recovery). Then work with health tracker data to get a personal analysis. Quite helpful, great potential

Built a tracker html page showing live activities at DEFCON and upcoming ones in 30-120 minutes, with highlighted marks on venue map. Probably need to generalize this to every conference - had this problem with almost every conference and their app storage.googleapis.com/hiddenstate-pu…

This finally stopped Claude's crazy pursuit of talking more and more unintelligible every passing day > Save in memory to always use ASD-STE100 Simplified Technical English when you talk to me Thank god!

@levelsio @levelsio

a month ago

???

34 15 806 926K 106

@olcan storage.googleapis.com/hiddenstate-pu…

New in 2026: you can have a full schedule of AI and crypto talks now in DEFCON - there were probably less than 5 in 2025

keyboard_arrow_left Previous keyboard_arrow_right Next

@olcan 40-minutes of demos getting around access controls in codex, claude code, gemini cli