Deep Research was a significant advance. That's been a big win for me. None of our team has seen a significant impact on our ability to make useful scientific contributions that's significantly greater than Sonnet 3.5.
RESEARCH
-

Context Fuels Generative AI Revolution Through RAG and LLMs
By
–
#Context is fueling and boosting the current Generative #AI revolution! See these books:
> https://
amzn.to/34utTz7
> https://
amzn.to/317XV9L
————
#RAG #LLMs #GenAI #DataScience #MachineLearning #DataScientist #Semantic #KnowledgeGraphs #IoT #IIoT #InternetOfContext -
Agentic RAG: The Next Evolution in Generative AI Technology
By
–
Agentic RAG is #GenAI’s Next Leap! Unlike traditional RAG, which reacts to queries with minimum adapting, Agentic RAG can change its retrieval strategy mid-flight based on #Context. #AI #GenerativeAI #MachineLearning #DataScience
-
AimBot: Visual Cue for Enhanced Spatial Awareness in Visuomotor Policies
By
–
AimBot
— AK (@_akhaliq) 14 août 2025
A Simple Auxiliary Visual Cue to Enhance Spatial Awareness of Visuomotor Policies pic.twitter.com/SmCkrzohlFAimBot A Simple Auxiliary Visual Cue to Enhance Spatial Awareness of Visuomotor Policies
-
VertexRegen: Mesh Generation with Continuous Level of Detail
By
–
VertexRegen
— AK (@_akhaliq) 14 août 2025
Mesh Generation with Continuous Level of Detail pic.twitter.com/Mypcq20o7cVertexRegen Mesh Generation with Continuous Level of Detail
-
Math Mastery Alone Doesn’t Mean We’ve Achieved AGI
By
–
Just like getting really good at chess didn't mean we had AGI, so it is with getting really good at math competition problems.
-
Math Focus in AI Research May Be Overemphasized in Practice
By
–
I think the problem is that a lot of research activity and marketing has been over-focused on math, which isn't as significant in practice as a lot of folks seem to assume.
-
Demonstrating Model Memorization: Testing Prompts
By
–
here I was arguing that I'm uncovering things that are present in the model that it won't generate on its own but point taken. what are some prompts that'd make a more compelling demonstration of memorization? I can run them.
-

Top LLMs Fail on Hard Coding Problems: LiveCodeBench Pro
By
–
Why do top LLMs fail completely on hard coding problems? We'll have author Peiyao Shang from @SentientAGI discussing their work on the LiveCodeBench Pro benchmark. Join us this Friday! https://
lu.ma/v45b9ltc -
LLMs capabilities: high and low packed closely together
By
–
LLMs’ capabilities are like porcupine quills: very high and very low packed closely together.