i'm a few days late to realizing this but: wow, opus 4.7 is god awful like so, so bad it's making mistakes on things i'd expect gpt-4o to handle cleanly there's got to be some explanation, right?
RESEARCH
-
GPT-5.5 achieves SOTA performance on long-running tasks
By
–
SOTA perf from GPT-5.5 for long-running tasks:
-
Open Benchmark Grants Initiative Funds Frontier AI Research
By
–
We're proud to have supported this work through Open Benchmark Grants, our initiative funding the next wave of frontier benchmarks!
-
GPT-5.5 Released: Major Performance Improvements Available
By
–
gpt-5.5 is a big step up in performance, give it a try:
-
Are LLMs Really More Important Than Fire or Electricity?
By
–
Are LLMs really more important than fire or electricity? “Honestly, a ton of what we’ve developed in my lifetime amounts to scaling up the delivery of information and entertainment and the frictionlessness of certain financial transactions. These are real improvements! … But
-
Token Importance in On-Policy Distillation with Selective Training
By
–
"TIP: Token Importance in On-Policy Distillation" This paper introduces selective token training for on-policy distillation, relying on student entropy to find high-signal tokens. A key point is that entropy misses confident mistakes, so they add teacher-student divergence to
-
Scaling Self-Play with Self-Guidance for Theorem Proving
By
–
"Scaling Self-Play with Self-Guidance" The main problem with self-play for theorem proving is that generator usually learns to reward hack, which produces messy hard problems that do not help the solver. This paper suggests by adding a Guide model that scores generated
-
AI Researchers’ Perspective on Existential Risk
By
–
Real AI researchers don't worry about AI extinction.
-
Top 5 AI News: GPT-5.5 and ChatGPT Images 2.0 Released
By
–
These are the top 5 AI news stories from the past week that you should actually care about: – @OpenAI dropped GPT-5.5 that’s supposed to need less prompting for better answers
– OpenAI also released ChatGPT Images 2.0 with people saying it beats Nano Banana
– @AnthropicAI