I feel my main velocity limitation lately isn't token speed anymore, it's compute. Running tests in parallel is taxing; can't wait for better cloud worker integration.

Exclusive: DoorDash launched a new app “Tasks” that pays couriers in some US markets if they submit audio and video clips to help improve AI and robotics models. Many of these tasks are completing household chores while capturing footage with a body-worn camera — data that would

🚨MAJOR BREAKING THE GOVERNMENT OF CANADA HAS JUST TABLED A NEW INTERNET MONITORING BILL which requires electronic service providers retain "prescribed metadata" of user activity for at least one year Bill text not released yet.

GPT-5.4 also has a 1M context window, but their evals show that needle-in-a-haystack (MRCR v2) scores 97% at 16-32K tokens, drops to 57% at 256-512K, and just 36% at 512K-1M. So it's a good idea to compact regularly!

GPT-5.3 Instant in ChatGPT is now rolling out to everyone. More accurate, less cringe. https://t.co/oJpXsp9TBc

excited for GPT-5.3 Instant rolling out today! a lot of work went into improving the everyday chatgpt experience, things that don’t always show up in benchmarks: better tone, fewer unnecessary refusals, and stronger answers from search. we’ve also reduced hallucinations ↓

GPT-5.3-chat shows a surprising & severe regression on EQ-Bench and Longform Writing. Lots of partial refusals on EQ-Bench. In the writing evals, the prose devolves to tiny 1-5 word paragraphs.

Most people in tech know Anthropic commands the majority of API spend by U.S. businesses. As of January, Anthropic took >50% of spend on enterprise AI subscriptions too. OpenAI still leads on business count. But the biggest spenders go to Anthropic. All from Ramp data.