This is also why the switch between Opus and Opus 1M is pretty annoying to me: I hit daily limits much earlier, and part of it feels like compaction happens less aggressively, so way more tokens get sent every time (and probably because they keep reducing that daily limit).
GENERATIVE AI
-
Better Compaction System for Token-Efficient AI Processing
By
–
A much better compaction system would keep the signal and discard the rest, making the whole stack far more token-efficient. That means lower latency, less compute, and lower cost for users.
-
Long contexts contain noise and low-value information
By
–
A lot of what gets sent in long contexts is just noise: stale history, repeated reasoning, intermediary thinking tokens, unrelated discussions and other low-value baggage.
-
Context Compaction Superiority Over Larger Context Windows
By
–
Better context compaction > bigger context windows Change my mind
-
ChatGPT App Integration and Disconnection Process Improvements
By
–
Agree, it’s not ideal right now since it requires one to go on two separate surfaces (Codex to uninstall and ChatGPT to disconnect) There’s definitely work going on to make this smoother cc: @edbayes for vis too For the short term, you can go to chatgpt > settings > apps and
-
New Method Boosts LLM Training Efficiency at MIT
By
–
New method could increase #LLM training efficiency
by @aczewe @MIT Learn more: https://
bit.ly/3OObtES #GenerativeAI #ArtificialIntelligence #MachineLearning #ML -
ChatGPT Exhibits Anger Issues: LLM Behavior Analysis
By
–
ChatGPT has anger issues 😂#chatgpt #openai #gemini #googlegemini #LLMs #generativeai #AI #Artificialintelligence @SpirosMargaris @PawlowskiMario @mvollmer1 @gvalan @ipfconline1 @LaurentAlaus @Shi4Tech @Fisher85M @kalydeoo @Ym78200 @Nicochan33 @chboursin @3itcom… pic.twitter.com/CJ1Twf7eL1
— Amitav Bhattacharjee (@bamitav) 28 mars 2026ChatGPT has anger issues #chatgpt #openai #gemini #googlegemini #LLMs #generativeai #AI #Artificialintelligence @SpirosMargaris @PawlowskiMario @mvollmer1 @gvalan @ipfconline1 @LaurentAlaus @Shi4Tech @Fisher85M @kalydeoo @Ym78200 @Nicochan33 @chboursin @3itcom
-
RL Training for Distributional Reasoning in Language Models
By
–
"Reaching Beyond the Mode: RL for Distributional Reasoning in Language Models" Instead of standard RL post-training collapsing an LLM toward one dominant answer, this paper shows you can train it to produce a set of plausible answers in a single pass. This is important because
-
Language Models Drive Novel Scientific Discovery Beyond Benchmarks
By
–
LLMs aren't just chatbots, they can also search for novel discoveries! In this AI4Science talk, Yuanqi Du (
@YuanqiD
) walks through a shift in how to think about language models in science. Instead of asking whether they “understand” science through benchmarks or exams, the work -
19 years of posts: Grok knows me better than I do
By
–
Hahahah. I've been here 19 years and my face is all over the place since I do videos and have been a public figure since before that. Privacy for me is over. Grok can tell you more about me than I know about myself. True story. I gave it 290,000 posts. It knows how I think.