https://t.co/rGFcIIZ7RV

Windows developers can now publish apps to Microsoft’s store without fees. Microsoft has removed the one-time fees for its store, and developers no longer need a credit card. Details 👇

Dear @zoox it's all awesome but when will your backyard in Foster City get a taste of Autonomous cars? We want Zoox first here in Foster City!

i'm hiring for a new team @openai: Applied Evals our goal is to build the world's best evals for the economically valuable tasks our customers care about most. we'll execute as a group of high‑taste engineers, combining hands-on, unscalable efforts with systems that others can

Summary of @elonmusk’s 44 minute interview at the @allinsummit: Government: • Elon hasn’t been back to DC since May, “The Government is basically unfixable.” • “If AI & Robots don’t solve the National Debt, we’re toast.” Optimus: • Tesla is finalizing the design of Optimus

A big part of our mission at Thinking Machines is to improve people’s scientific understanding of AI and work with the broader research community. Introducing Connectionism today to share some of our scientific insights.

Updated & turned my Big LLM Architecture Comparison article into a narrated video lecture. The 11 LLM architectures covered in this video: 1. DeepSeek V3/R1 2. OLMo 2 3. Gemma 3 4. Mistral Small 3.1 5. Llama 4 6. Qwen3 7. SmolLM3 8. Kimi 2 9. GPT-OSS 10. Grok 2.5 11. GLM-4.5

Quite a new perspective

Top AI Papers of The Week (September 1-7): - rStar2-Agent - Self-Evolving Agents - Adaptive LLM Routing - Universal Deep Research - Implicit Reasoning in LLMs - Why Language Models Hallucinate - Limitations of Embedding-based Retrieval Read on for more:

building pretraining infrastructure is an exercise in complexity management, abstraction design, operability/observability, and deep systems and ML understanding. reflects some of the trickiest and most rewarding problems in software engineering. which makes it really fun!

Reminds me of one old joke... Yeah, see, I can type really, really fast, but it's all garbage.

You can’t build AI products like other products, for two reasons: 1. AI products are inherently non-deterministic 2. Every AI product must negotiate a tradeoff between agency and control When companies don’t recognize these differences, their AI products face unexpected

The trick is to correct binary classifiers. Sounds easy. But is it?

Coming later today: what @sama and Jony Ive are up to. As learned this morning in the streets of San Francisco from a venture capitalist who is already investing in companies that are building for it. The strategy that will decrease Apple’s relevance to all our lives.

xAI Seattle office opening

First, Atlassian started throttling @glean (in June: https://t.co/YeVp93LYLI) Now, they've acquired @browsercompany. What's next? I won't be surprised if Atlassian started building it's own merge of AI Agents with its Jira/Trello offering, similar to @useMotion. Agentic wars!

Agentic wars? First, Atlassian started throttling @Glean ( https://t.co/YeVp93LYLI). Second, Atlassian acquired independent agentic browser. Wonder if next step is transforming Jira/Trello into something like @useMotion with search across all team's data & AI agents inside?

You can now run 100B parameter models on your local CPU without GPUs. Microsoft finally open-sourced their 1-bit LLM inference framework called bitnet.cpp: > 6.17x faster inference > 82.2% less energy on CPUs > Supports Llama3, Falcon3, and BitNet models

My team at @Apple is looking for ML researchers/engineers in Multimodal LLMs for Human Understanding. In this role, you can both publish your research and ship your ideas to impact billions of users. We focus on topics like VLMs, RL-finetuning, Pretraining Streaming Video

Awesome. At last!

A Microsoft employee has built a driver that restores life to abandoned Windows Mixed Reality headsets with SteamVR on Windows 11 version 24H2