building models is a juggling act, we optimized 05-06 for certain code tasks at the cost of other tasks, good lessons learned that the small regressions we saw in evals tracked to how people felt about the model, but we closed the gaps with 06-05!
RESEARCH
-
Sam Altman OpenAI Drama Captivates AI News Watchers
By
–
Haven’t been this glued to the newsfeed since Sam was kicked out of OpenAI.
-
Stanford Researchers Improve Domain Generalization AI Benchmarks
By
–
With support from @StanfordHAI
, a team of Stanford researchers investigates when and why current domain generalization benchmarks fall short and how they might be improved: https://
stanford.io/4kTMQl3 -

Gemini 2.5 Pro Upgrade Achieves SOTA Benchmarks
By
–
Now on Poe: Gemini 2.5 Pro just got an upgrade that is crushing benchmarks (SOTA on HLE, Aider, GPQA, and more). Key improvements include enhanced multi-turn conversation abilities, complex reasoning, sophisticated coding, and creative writing. (1/2)
-
Widespread Industry Practice at Major Tech Companies Revealed
By
–
Was Google the only company at which that happened? It was not, and *everyone in tech knows it.*
-
Quality Crisis in AI-Generated Content Over Recent Years
By
–
I do not think that people appreciate how many gigabytes of Holy Shit You Put That In Writing And Think You Deserve A Gold Star For Doing So were authored the last few years.
-
Multiple Global Crises Converging: AI Doom Scenario Discussion
By
–
So we are really gonna get Great Depression II, Civil War II, World War III, and the AI Doom all at the same time, aren’t we?
-
AI Breakthroughs Accelerating Weekly Toward 2040 Singularity
By
–
Everyone who is following AI news noticed the trend: every week we get one or more monumental advances. Each would've been world-shattering a decade ago.
Keeping up is nearly a full-time job! Now, imagine 2040, as we near the technological singularity: 2040, Week 1: