Your curated collection of saved posts and media
π¨ New: We built @a16z's personal GPU AI Workstation Founders Edition - 4x NVIDIA RTX 6000 PRO Blackwell Max-Q (384GB total VRAM) - 8TB of NVMe PCIe 5.0 storage - AMD Threadripper PRO 7975WX (32 cores, 64 threads) - 256GB ECC DDR5 RAM - 1650Watts at peak (runs on a standard 15Amp/120V circuit). For training, AI research, and deploying models locally. A datacenter-class AI rig you can keep under your desk. We are planning to make a limited number of these a16z AI Workstations. Build guide + how you can make your own π
xAI finally open sources Grok 2 https://t.co/5ljUvALefJ
@BasedBeffJezos Itβs high time we open sourced Grok 2. Will make it happen next week. Weβve just been fighting fires and burning the 4am oil nonstop for a while now.
link: https://t.co/BWzNygIupW
Anyone else remember this tagline? My dream for Kaggle was to help make the best ML experts big names with big pay-packets, like sport stars. It's finally happening -- the world is starting to realise the value of the best people in this field is huge. https://t.co/DigpwzHg98
Grok 2 from @xai has just been released on @huggingface: https://t.co/vT2CX3dgud https://t.co/OLVciB1gAA

Grok 2 from @xai has just been released on @huggingface: https://t.co/vT2CX3dgud https://t.co/OLVciB1gAA

Grok2 is open source now and available on Hugging Face. I have 2 questions: - wtf is `model_type: doge` - wtf is this rope theta value https://t.co/cJ4mu6UDad
.@xaiβs Grok 2 weights have been released on @huggingface https://t.co/KLPTNA8JZ1 https://t.co/IheMMjnbW3

.@xaiβs Grok 2 weights have been released on @huggingface https://t.co/KLPTNA8JZ1 https://t.co/IheMMjnbW3

Hunyuan 3D-2.1 turns any flat image into studio-quality 3D models. And you can do it on this @huggingface space for free. https://t.co/NYxRmQ2mUJ
https://t.co/9CNCEyBbFA
https://t.co/9CNCEyBbFA
Runway has a new game engine thing that creates interactive comics with choices in any style you want. This is what it did when I gave it the prompt "explore noses." Similar, but more graphical than previous AI Dungeon-like work, with some decent systems to keep things on rails. https://t.co/768TAIBECS

Car colors have gotten boring thanks to a feedback loop: a plurality has always liked boring colors β‘οΈdealers stock those colors β‘οΈyou have to wait for to get a cool color β‘οΈless people get cool colorsβ‘οΈcool colors seem weird to more consumers π Charts for America & Poland. 1/2 https://t.co/YPyS2324qI

One interesting way to understand or fact-check claims is to ask an AI to make a simulation of a process and see if it makes sense. "Gemini 2.5 (with Canvas on), create a simulation showing this feedback loop''' https://t.co/Y6GmqhaFRg
Car colors have gotten boring thanks to a feedback loop: a plurality has always liked boring colors β‘οΈdealers stock those colors β‘οΈyou have to wait for to get a cool color β‘οΈless people get cool colorsβ‘οΈcool colors seem weird to more consumers π Charts for America & Poland.
In an earlier paper, re-insurer Swiss Re looked at third party claims & found Waymo led to "a 9-times reduction in property damage and 12-times reduction in bodily injury claims to the overall on-road vehicles in the areas that the Waymo ADS currently operates (the status quo)" https://t.co/ZKeXvbQlng

Agents learned how to gamble. Casino season is coming up soon to @crestalnetwork https://t.co/fWHBOWw81w

InternS1 Tech report is now available on @HuggingPapers π₯ https://t.co/pCkeXlRlcX https://t.co/UQNNbizx9B
Intern-S1-mini π₯ lightweight open multimodal reasoning model by @intern_lm https://t.co/aTXFjyv6Q5 β¨ Efficient 8B LLM + 0.3B vision encoder β¨ Apache 2.0 β¨ 5T multimodal pretraining, 50%+ in scientific domains β¨ Dynamic tokenizer for molecules & protein sequences

InternS1 Tech report is now available on @HuggingPapers π₯ https://t.co/pCkeXlRlcX https://t.co/UQNNbizx9B

AWorld introduces a dynamic multi-agent system that achieved 1st place on the GAIA leaderboard. @_akhaliq @Grad62304977 Check out the paperοΌ https://t.co/zfQmJ22luF
Intern-S1 A Scientific Multimodal Foundation Model https://t.co/vRrRWl1z6y
discuss with author: https://t.co/aST05WrmBt
KLING is BEST on Higgsfield. KING is in the building! SMARTEST Start & End frames with the HIGHEST quality of 1080p. High IQ Transitions & High Aesthetic Morphs. Retweet to get the FULL Kling guide in your DMs. THE King is here to serve you. Let it. https://t.co/hgWYe2n47k
Deep Think with Confidence https://t.co/Owq8C89INU
discuss with author: https://t.co/BCxgNrlWIH
Mobile-Agent-v3 Foundamental Agents for GUI Automation https://t.co/R1N6DStyrH
discuss with author: https://t.co/KtuhwB5DGG
LiveMCP-101 Stress Testing and Diagnosing MCP-enabled Agents on Challenging Queries https://t.co/Lt03rX75Hg
discuss with author: https://t.co/hmrHWsbE0j
text to video for vibe coded sites https://t.co/YkTlpZPQeh
We dropped 30+ reasons to break the internet. New viral presets for the WAN 2.2 model are live on Higgsfield. Ahegao, Abs Show Off, Jojo poses, new camera moves & more. The fastest model on earth, now with one-click results. Retweet to get the full guide in DM. https://t.co/E8YnIEy6Ru