Gemini 3.5 Flash outperforms 3.1 Pro on many vision use cases (like the below Roboflow eval) while being ~6x faster on average Gemini multimodal understanding for the win.
RESEARCH
-
DeepSeek v4 Pro 75% Discount, Only 27% Compute and 10% Cache
By
–
Let that sink in for a moment. DeepSeek v4 pro 75% discount. Permanent! In: $0.43
Out: $0.87 If you read the DeepSeek v4 tech paper you know that this model is insanely good when it comes to efficiency. Only 27% compute and only 10% cache compares to v3.2. SemiAnalysis wrote -
Supercomputing for AI: foundations, architectures, scaling deep learning
By
–
Supercomputing for AI — Foundations, Architectures, and Scaling Deep Learning. [804-page masterpiece] Get it: https://
amzn.to/4qS4pFz Git it: https://
github.com/jorditorresBCN
/supercomputing-for-ai
… -
Principles of Deep Learning Theory PDF and Book Edition
By
–
Principles of Deep Learning Theory—Theoretical & Mathematical Foundations
[471-page PDF Draft] http://
arxiv.org/abs/2106.10165
-or-
Buy new edition: http://
amzn.to/3qoqmS5
———————
#DataScience #AI #MachineLearning #ML #NeuralNetworks #Algorithms #Mathematics #DataScientist -
Machine Learning Refined: Foundations, Algorithms, Applications with Python
By
–
Machine Learning Refined — Foundations, Algorithms, and Applications (with 100 in-depth coding exercises in Python): http://
amzn.to/3EblXVx
—————
#DataScience #ML #AI #Mathematics #DataScientist -
Open Source Deep Research for Agent Harnesses with AI-Q Skills
By
–
Say hello to open source deep research for your favorite agent harness.
— NVIDIA AI (@NVIDIAAI) 22 mai 2026
Our AI-Q agent skill packages the work of building a research pipeline into a portable skill. Drop it into your harness, and the agent delegates a research task to a local or hosted AI-Q server and gets back… pic.twitter.com/u7tVWvCUWdSay hello to open source deep research for your favorite agent harness. Our AI-Q agent skill packages the work of building a research pipeline into a portable skill. Drop it into your harness, and the agent delegates a research task to a local or hosted AI-Q server and gets back
-
Autonomous Preference Optimization transforms AI model disagreements into constraints
By
–
What if shifting reasoning patterns from multiple AI models could be turned into constraints instead of noise? Researchers from UTS’s Australian AI Institute (AAII) introduce Autonomous Preference Optimization (APO). Their approach treats disagreements between models as dynamic
-
TRINITY: 0.6B Model Learns to Manage Multiple Tasks
By
–
A 0.6B model learned to manage giants. That is the idea behind TRINITY, a new ICLR 2026 paper by Jinglue Xu, Qi Sun, Peter Schwendeman, Stefan Nielsen, Edoardo Cetin, and Yujin Tang. The paper is not asking: “How do we build one model that knows everything?” It is asking
-
TRINITY: 0.6B Model Managing Giants (ICLR 2026)
By
–
A 0.6B model learned to manage giants. That is the idea behind TRINITY, a new ICLR 2026 paper by Jinglue Xu, Qi Sun, Peter Schwendeman, Stefan Nielsen, Edoardo Cetin, and Yujin Tang. The paper is not asking: “How do we build one model that knows everything?” It is asking
-
Codex Publishes Model Training Experiment Blogpost
By
–
codex is writing a blogpost about its experiments in training a model all by itself