Found my first clear GPT-4.5 winner task: asking it about obscure star wars facts ("do you remember ar'krai from the new jedi order") causes other models to hallucinate. Only 4.5 actually gets it right. Anyone else have similar obscure queries?

Today we're thrilled to highlight Bobby Youstra, a Computer Science at UT Austin graduate pursuing a career in the game industry! This summer, he'll be joining Epic Games as an Engine Programmer Intern, and we can't wait to see all he accomplishes there. Congratulations, Bobby! #GameDev #UTAustin

biggest advantage of long-context ai is "here's 1M+ tokens of contrastive logs, something weird happened, find me why it happened"

cerebras.ai/blog/introduci… Cerebras:

For the '30 hours of autonomous coding' that's being thrown around for claude 4.5, I wonder how much was spent waiting on bash to return? I see it using at times hourlong timeouts. Still very cool even if so, but just a wide, wide gap between these two worlds

We’re opening applications for the next two rounds of the Anthropic Fellows Program, beginning in May and July 2026. We provide funding, compute, and direct mentorship to researchers and engineers to work on real safety and security projects for four months.

A: "Nvidia has been known to do some funky numerical things. You know that time when the result looked like fp32 but they secretly accumulated in 22 bits? A lot of folks got mad at that one" B: "Oh you mean the tf32 dtype, which is actually secretly fp23?" C: "No, I'm talking