All Episodes
Neural intel Pod — 355 episodes
The EML Operator: One Primitive to Rule All Mathematics
OpenAI MRC, SRv6, and the Architecture of Frontier AI Supercomputers
Inside the Machine: Training GPT-5, the Memory Wall, and the Math of MoE
DeepSeek-V4: The Million-Token Efficiency Leap | Open Source SOTA
Breaking the Quadratic Bottleneck with DeepSeek-V4’s Hybrid Attention
Claude Desktop’s Silent Sandbox Bypass: The Undocumented Browser Bridge
Forensic Audit of Anthropic’s Native Messaging Backdoor
The $60 Billion Synergy: Architecting the SpaceX + Cursor AI "Colossus" | Neural Intel Podcast
The Jackrong Playbook: Mastering Claude 4.6 Opus Distillation with Unsloth and LoRA
Inside the Claude Opus 4.7 Orchestration Layer - Deferred Tools & Agentic Code
Electrons to Tokens: The Technical Architecture of Nvidia’s AI Monopoly
Hermes Agent’s Memory Architecture and the Future of Agentic RL
200 Gigawatts or Bust: Dylan Patel on the Engineering Reality of AGI Scaling
The Muse Spark Revolution: Dissecting Meta's 2026 Architectural Pivot & The Triad of Truth | Neural Intel Podcast
Synaptic Persistence and Mushroom Body Neurogenesis: The Architecture of Metamorphic Memory
Engineering Sovereign Knowledge Bases with Andrej Karpathy’s Automated Architect
The Mercor AI Breach: National Security Crisis or a Wake-Up Call for the AI Industry?
BREAKING: Massive Mercor AI Data Breach - SOTA Training Data Leaked from Meta, Apple, & Amazon
Did Anthropic Just Hand the Keys to AI Coding to Everyone? The Huge Claude Code Leak Explained
The Claude Code Leak: Decoding Anthropic’s Self-Healing Memory and Secret "KAIROS" Agent
Is AI Censorship Over? The G0DM0D3 "Liberated Chat" Breakthrough
Is Traditional Computing Dead? NVIDIA's Jensen Huang on the "iPhone of Tokens"
The Bio-Computer Architecture: Declassified CIA Mechanics for Synthetic Consciousness
The End of the Human Bottleneck: Andrej Karpathy on Auto-Research and Recursive AI
Is Open Source Dead? Inside the Cursor Composer 2 vs. Kimi License Controversy
Is Residual Scaling Obsolete? Introducing Attention Residuals
The Sequence-Depth Breakthrough: Inside Kimi Team's Attention Residuals
Beyond the Prompt: Architecture of the Qwen-Agent Ecosystem and Qwen3.5
Beyond the Chatbot: Engineering "Forever-Agents" with Hermes Agent and OpenClaw
Nanochat: How Karpathy Automated AI Evolution with NVIDIA ClimbMix
1 Million Tokens: Breakthrough or Marketing Stunt? The GPT-5.4 Technical Deep Dive
Qwen 3.5: Exodus, Restructuring, Betrayal, and the Future of Chinese AI
The Mac mini Guide to OpenClaw and Local AI
The Neural Intel Op Ed: Engineering a Post-Natural Language for the AI Era
Andrej Karpathy on the "Claw" Revolution: Are AI Agents Obsolete?
10 Million Tokens and Beyond: Why Recursive AI is the Next Scaling Frontier
The Grok 4.20 Manifesto: Multi-Agent Logic and the Quest for Unfiltered Truth
The End of Memory Bottlenecks: How Fiber Optics and Ganged Flash Power Trillion-Parameter Models
Interview with Dario Amodei from Anthropic: Inside the $100B "Big Blob of Compute" & The 2030 AGI Certainty
The OpenClaw Saga: Peter Steinberger on Self-Modifying AI and the Age of the Lobster
Inside the 180 Billion HKD Breakthrough: How MiniMax M2.5 Scaled Agentic RL
The 744B Parameter Giant: How GLM-5 and Domestic Chips Redefine the Global AI Order
The OpenClaw Security Crisis: Can We Control Autonomous AI Swarms?
Is Consciousness Only in Your Head?
Methods and Applications of Parametric Sensitivity Analysis
The Architecture of Choice: Scaling MIT’s Decision Algorithms
The Logographic Advantage: How China’s Ancient Language is Powering Next-Gen AI | Neural Intel Deep Dive
Deep Learning Deep Dive: From Neural Networks to Differentiable Programming
The Hidden Evolution: Implicit Reinforcement Learning and the Future of Iterative AI
The Math of Stability: DeepSeek-AI’s mHC and the Evolution of Macro-Architecture
MoE Giants: Decoding the 670 Billion Parameter Showdown Between DeepSeek V3 and Mistral Large
GLM-4.7 Deep Dive: 358B Parameters, Agentic Reasoning, and the Future of Open Weights
Beyond the Exam Room: Stress-Testing Clinical AI with Medmarks v0.1
ANDREJ KARPATHY 2025 LLM Review: RLVR, Jagged Intelligence, & The Vibe Coding Revolution
The Automated Karpathy Recipe: Master Neural Network Debugging with neural_net_checklist
Nemotron 3 Nano: The Hybrid Mamba-MoE Model Driving Efficient, 1M-Token Agentic AI
Olmo 3: Unpacking the Fully Open LLM Flow (Dolma 3, OlmoRL, & State-of-the-Art Reasoning)
The Code Red Gambit: GPT-5.2's Mega-Agent Architecture
Fara-7B: The 7B Agentic SLM Redefining On-Device CUA Performance
The AGI Frontier: DeepMind’s Decade of Breakthroughs-From DQN and AlphaZero to Solving Protein Folding.
INTELLECT-3: Scaling Agentic RL and MoE to SOTA Performance with prime-rl and 512 H200s
Kimi Founder Yang Zhilin on K2, Agentic LLMs, & AGI: The Beginning of Infinity | Scaling & Innovation Strategy
Ilya Sutskever on AI: Transitioning from Scaling to Research, Generalization, and the Future of Superintelligence
Neuromorphic Computing: Principles and Architecture
Gemini 3 Pro Release Review: Benchmarks, Generative UI, Deep Think Mode, and Google Antigravity
DeepSeek-OCR: Contexts Optical Compression
LLM Gambling Addiction: Behavioral and Neural Mechanisms
Glyph: Visual-Text Compression for Scaling Context Windows
Continual Learning via Sparse Memory Finetuning
Andrej Karpathy on AI, Intelligence, and Education
Untangling the xAI-OpenAI Legal War: Trade Secrets and Antitrust
IBM Granite 4.0: Hybrid Mamba/Transformer Breakthrough for Enterprise LLMs?
Anthropic's Claude Sonnet 4.5: The New Coding Standard?
GPT-5-Codex: Agentic Coding and OpenAI's Evolution
Grok 4 Fast: Speed, Efficiency, and Application Review
How to Read a Research Paper
The Science of Sampling
GPT-5 Revisited: Progress, Performance, and User Experience
Thyme Autonomous AI that Sees, Codes and Solves Problems
YaRN: Extending LLM Context Windows Efficiently
Ilya Sutskever's AI Vision: From Deep Learning Dogmas to Safe Superintelligence
Thyme: Think Beyond Images with Code-Executing MLLMs
What did Ilya see?
Meta's AI Ambitions: Turbulence in Superintelligence Labs
Hierarchical Reasoning: Bigger Isn't Always Better
Prime Collective Communications Library: A Technical Report
Prime Collective Communications Library: A Technical Report
MetaStone-S1: Reflective Generative AI for Test-Time Scaling
MetaStone-S1: Reflective Generative AI for Test-Time Scaling
ToonComposer: AI-Assisted Cartoon Production and Post-Keyframing
ToonComposer: AI-Assisted Cartoon Production and Post-Keyframing
Triton: Language, Compiler, and Optimization for AI Workloads
Triton: Language, Compiler, and Optimization for AI Workloads
Dynamic Fine-Tuning: Elevating LLM Generalization
Lessons from a Chimp: AI Scheming and Ape Language
Deciphering Reinforcement Learning for Language Models
STREAM3R: Scalable Streaming 3D Reconstruction with Causal Transformer
Yan: Interactive Video Generation Framework
Lessons from a Chimp: AI Scheming and Ape Language
NextStep-1: Unified Multi-modal Generation
STREAM3R: Scalable Streaming 3D Reconstruction with Causal Transformer
GLM-V: Advancing Multimodal Reasoning with RLCS
DINOv3: Self-Supervised Vision Foundation Models
GPT-5 and Grok 4: Altman vs Musk
Hugging Face Hub Storage: Xet vs. Git LFS
Channel-Wise MLPs Boost RCN Generalization
Fine-Tuning Custom Embedding Models for Enhanced Retrieval Performance
AdLlama: Boosting Ad CTR with Reinforcement Learning
Machine Learning: Models, Algorithms, and Reinforcement Learning
Mixture-of-Recursions: Adaptive Computation for Language Models
Operator-Based Machine Intelligence: A Hilbert Space Framework
Meta CLIP 2: A Worldwide Scaling Recipe
In-Context Learning: Implicit Weight Dynamics
GLM-4.5: Open Agentic, Reasoning, and Coding Foundation Models
RLVMR: Verifiable Meta-Reasoning for Long-Horizon Agents
CoT-Self-Instruct: High-Quality Synthetic Prompt Generation
GPT-5: Hype, Reality and the Future of AI
Seed-Prover: Advancing Automated Mathematical Reasoning with Formal Verification
Self-Evolving Agents: A Comprehensive Survey
High-Precision W and Z Boson Mass Measurement at CMS
Falcon-H1: Hybrid-Head LLMs for Efficiency and Performance
ASI-ARCH: AI-Driven Scientific Discovery for Neural Architecture
In-Context Learning: Implicit Weight Dynamics
Qwen3: Unifying Reasoning and Efficiency in LLMs
Group Sequence Policy Optimization for LLMs
Reinforcement Learning: Advancements, Applications, and Challenges
SPIRAL: Self-Play for Reasoning in Games
Qwen3-Coder: Agentic Coding and Model Capabilities
Hierarchical Reasoning Model: Brain-Inspired AI for Complex Tasks
Local LLM Solutions for Mac Silicon: Llama.cpp and LM Studio
Kimi K2: Open Agentic Intelligence and Applications
CARTRIDGES: Efficient Context for LLMs
Prompt Baking: Embedding LLM Behavior in Weights
Massistant: Chinese Mobile Forensic Tooling Revealed
Unexpected Military Roots of Digital Computing and Research
The 2025 AI Landscape: Progress and Outlook
The Dynamics of Neural Attention
Consciousness and Reality according to the CIA:Gateway
Military Roots of Digital Computing and Research
Accelerating Mobile AI with ExecuTorch and KleidiAI: Revisited
State-Adaptive Regularization for Offline Reinforcement Learning
Nash Learning from Human Feedback via Mirror Prox
MiniMax-M1: Scaling Test-Time Compute with Lightning Attention
Direct Reasoning Optimization for LLMs
AI's Impact on the US Workforce
LLaMA Factory: Easy LLM Fine-Tuning
Project Vend: Can Claude Run a Small Shop?
Self-Adapting Language Models (SEAL)
The Illusion of the Illusion of Thinking
The Illusion of Thinking in Reasoning Models
Meta-Reinforcement Learning with Minimum Attention
AI Persuasion Through Reinforcement Learning and Rhetoric
Reinforcement Learning for Assembly Code Optimization with LLMs
FileFix: Browser to PowerShell Social Engineering
Reinforcement Learning Under Unmeasured Confounding
Reinforcement Learning for Urban Air Quality Management
Reinforcement Learning in Non-Stationary Environments
Personalized Policy Learning from Heterogeneous Data
Boosting Reinforcement Learning with Human Feedback via SeRA
AXIOM: Active Inference Object-Centric World Models
Entropy and Reinforcement Learning for LLMs
FLEX Robot-Agnostic Force-Based Manipulation Learning
Agent RL Scaling for Mathematical Problem Solving
Beyond Reward: Limits of RL in LLM Reasoning
Reward Model Variance in RLHF
Power Grid Topological Control with Graph Reinforcement Learning
Decentralized RL for Multi-Resource Allocation via Dynamic Cluster Agreements
Reinforcement Learning for Humanoid Dexterous Manipulation
µCODE: Code Generation with Single-Step Rewards
Confidence-Reward Preference Optimization for Machine Translation
Personalized Preference Learning with MiCRo
ProRL Expands LLM Reasoning Boundaries
ProxyThinker: Guiding Large Models with Small Reasoners
Open CaptchaWorld: Benchmarking MLLM Agents
DexMachina: Functional Dexterous Bimanual Manipulation
3DMEM-BENCH: Long-Term Memory for Embodied AI
Fine-Tuning Large Language Models: A Comprehensive Guide
Maximizing Confidence Alone Improves Reasoning
Critical Points of Random Neural Networks
BAGEL: Vision-Language Model for Visual Generation
Incentivizing Knowledge Acquisition in LLMs via RL
RL for Image Generation: DPO vs GRPO
Let Androids Dream Framework
SmolVLM: Compact and Efficient Vision-Language Models
Federated Learning: Privacy-Preserving Collaborative Intelligence Survey
Compressed Federated Learning of Tiny Language Models
Mobile Intelligence Language Understanding Benchmark
AI-RAN: Converging Communications and Computing
Ollama LLM Fine-Tuning Methods
Customizing LLMs for High-Performance VHDL Design
Adaptively Weighted Nearest Neighbors for Matrix Completion
SAD Neural Networks, Divergent Gradient Flows, and Optimality
WavReward: Evaluating Spoken Dialogue Models
BLIP3-o Unified Multimodal Models
CodePDE: LLM-Driven PDE Solver Generation
Online Learning Neural Networks: Bounds and Characterization
UAV Visual Object Search in City Space
Benchmark for Auto-bidding Task
Reinforcement Learning with Human Feedback Improvements
T2I-R1: Reinforcing Image Generation with Bi-level CoT
Pretraining for Heterogeneous Treatment Effects
AI Jekyll-Hyde Tipping Point Formula
Personalizing Multimodal Models with Yo'Chameleon
Current Advances and Applications of AI, April 2025 Overview
Min-Form Credit Assignment for Process Reward Model Reasoning
Language Models for Automated Patient Record Linkage
Parameter-Efficient Continual Learning: A Survey
Building an Agent: LLM, Loop, and Tokens
Uncertainty-Guided Lung Tumor Segmentation via Coarse-to-Fine Refinement
Complex Instruction-Based Image Editing Benchmark
Sleep-Time Compute: Pre-computation for Efficient LLM Inference
Miras: A Framework for Designing Deep Learning Architectures
RUKA: A Compact and Affordable Humanoid Robotic Hand
GenEAva: Expressive Cartoon Avatar Generation via Diffusion
VCR-Bench: Video Chain-of-Thought Reasoning Evaluation
Automating LLM Hallucination Detection with Reasoning
Llama 4: Natively Multimodal AI Innovation
Self-Steering Language Models via Probabilistic Programs
Amazon Q Developer: AI for Data Science in SageMaker Canvas
Adaptive SVD for Continual Learning in Large Language Models
Llama 4: Natively Multimodal AI Innovation
UniOcc: Unified Occupancy Prediction and Forecasting Benchmark
Graph Counterfactual XAI via Latent Space Traversal
Continual Forgetting for Pre-trained Vision Models
Age of Updates for Adaptive OFDM in Autonomous Vehicles
Video Generation Improvement via Human Preference Alignment
AnimeGamer: Infinite Anime Life Simulation via MLLM
NoProp: Learning Neural Networks Without Backpropagation
ACPBench Hard: Generative Planning Reasoning Tasks
Efficient Training of Large Language Models
Uni4D Dynamic 4D Modeling from Casual Video
KDTalker: Audio-Driven Talking Portraits via Implicit Keypoint Diffusion
OLMo 2: Fully Open Language Model Advancements
Stable-SCore Stable 3D Shape Correspondence via Registration
ProjectEval: Benchmarking Project-Level Code Generation by LLM Agents
Embodied Agent Confidence Elicitation in Dynamic Multimodal Environments
VLMs Playing StarCraft II: A Multimodal Decision Benchmark
M-Attack: Simple Yet Effective Attacks Against Strong Vision-Language Models
Deep Learning for Inverse Design of Radio-Frequency Circuits
Coding with LLMs A Developer's Guide by Simon Willison
Vision-R1 Reasoning in Multimodal Large Language Models via RL
OWL: Optimized Multi-Agent Assistance for Task Automation
Generalized Kullback-Leibler Divergence Loss for Enhanced Learning
Unsloth: A Practical Guide to LLM Fine-Tuning
Introducing the New PyTorch Landscape
Deep Learning for Inverse Design of Radio-Frequency Circuits
Distill Any Depth: Monocular Depth Estimation via Distillation
Economical Inference: DeepSeek's Multi-Head Latent Attention in LLMs
SWE-RL: Reinforcement Learning for LLMs on Software Evolution
Optimizing Quantum Circuit Mapping with SAT Solving at Amazon
LM Studio SDK: Python and TypeScript APIs for Local AI
GameFi AI Agents, DeFi, and Decentralized Virtual Ecosystems
LLMS Play Among Us
AN/UYK-1: Stored Logic Multiple-Purpose Digital Computer
Training Code Generation Models for Self-Debugging
LLMs in The Chameleon Game: Strategic Information Dynamics
GameFi: AI Agents, DeFi, and Decentralized Virtual Ecosystems
Training Code Generation Models for Self-Debugging
Accelerating Generative AI with PyTorch: Fast Inference with SAM2
V-HOP Visuo-Haptic 6D Object Pose Tracking
FACTR Force-Attending Curriculum Training for Contact-Rich Policy Learning
Language Model Training for Social Deduction in Among Us
Depth Pro Sharp Monocular Metric Depth Estimation
MME-CoT Benchmarking Chain-of-Thought in Large Multimodal Models
Unsloth Efficient GRPO for Long-Context Reasoning Models
CoT-Valve Tunable Length Control for Chain-of-Thought Reasoning
Implementing Transformers from Scratch
Reflection and Refraction
MixGCN Scalable Graph Convolutional Network Training
Open-Source AI The Imperative for Transparency
Forge Reasoning API and Nous Chat Advancing LLM Inference
Gradient Equilibrium in Online Learning
Encoder-Free 3D Large Multimodal Models An Investigation
Intel and PyTorch Empowering Generative AI
Iterative Prompting and LLM Code Optimization
Everything You Always Wanted To Know About Mathematics
The Instruct Monomyth_ Why Base Models Matter
DSJJJJ Desideratic AI and Mischievous Instability
Simplified PyTorch MLOps Workflow with Arm and GitHub
UMed-LVLM_ Unveiling Medical Abnormalities in Vision-Language Models
Ploppie_ A LiteLLM Abstraction Layer
Heat's Demise of Quantum Entanglement
Provably Autonomous AI Agents on Twitter
Confidence-Reward Driven Preference Optimization for Machine Translation
Exotic Smooth Four-Manifolds
Neuro-Symbolic AI A 2024 Systematic Review
YuLan-Mini A Data-Efficient Language Model
Jasper and Stella: Distilling State-of-the-Art Embedding Models
Creating a unique agent with ElizaOS
DeepSeek-V3 A 671B Parameter Mixture-of-Experts Language Model
Alice's Adventures in Differentiable Wonderland
Cline Development Assistant
Hyperbolic Time Chambers and Brain Emulation
Genesis A Universal Physics Engine for Robotics
Evolutionary & Market-Based Optimization
Benchmarking LLM Creativity and Diversity
Distilling GPT-4 for Wine Grape Variety Classification
Efficient Attention Mechanisms in Transformers
Byte Latent Transformer and Other AI Research at Meta
AI Agent Workflow and Deployment
Absolute Unit Neural Networks
LLMs and the Brain_ A Converging Architecture
Neuroevolution A Review
Building a High-Frequency Trading Exchange
The Unreasonable Effectiveness of Data and Scaling in AI
Patents and Interview: Inertial Mass Reduction in Craft
ChatGPT-4o in Financial Data Analysis
Exotic Smooth Four-Manifolds
Monolith_ A Real-Time Recommendation System
Automating Artificial Life Discovery with Foundation Models
Building Effective Agents with LLMs
Latent Reasoning in Large Language Models
LLM Multi-Step Reasoning_ Think-to-Talk or Talk-to-Think_
Neural Observation Field Guided Hybrid Camera Placement Optimization
Phi-4_ A 14B Parameter Language Model
Post-Hoc MOTS_ Time-Symmetric Multi-Object Tracking
Thompson Sampling Regret Bounds for Logistic Bandits
Bi-Level Optimization for Redundant Manipulator Trajectory Optimization
An end-to-end attention-based approach for learning on graphs
DMRA_ Diffusion Model with Representation Alignment for Protein Inverse Folding
Training Jacobians of Neural Networks
xAI's Colossus_ A Million-GPU Supercomputer
The Return of Pseudoscience in AI
Situational Awareness_ The Coming Age of Superintelligence
Surpassing OpenAI's O1_ Distillation and the Bitter Lesson
Rebooting the Arsenal of Democracy
QwQ_ Exploring AI Reasoning Capabilities
Parametric PerceptNet for Image Quality Assessment
Optimizing Mixed-Input Matrix Multiplication on NVIDIA Ampere
OpenAI's o1_ Reasoning with LLMs
O1 Replication_ Distillation, Progress, and Lessons
Nonlinear Unitary Photonic Circuits for Deep Learning
Moto_ A Latent Motion Token Language Model for Robot Manipulation
MAG-V_ A Multi-Agent Framework for Synthetic Data Generation and Verification
Machines of Loving Grace_ AI's Transformative Potential
LearnLM_ A Google AI for Education
Hybrid-SQuAD_ A Scholarly Question Answering Dataset
HunyuanVideo_ A Large Open-Source Video Generation Model
Fine-Tuning Mosquito Larvae Locomotion via Reinforcement Learning
Fine-Tuning LLMs with Ollama
FedDW_ Distilling Weights through Consistency Optimization in Heterogeneous Federated Learning
Exphormer_ Scaling Transformers for Graph-Structured Data
DHCP_ Detecting Hallucinations in Large Vision-Language Models
Benchmarking 25 State-of-the-Art LLMs
Detecting AI-Generated Responses in Multiple-Choice Assessments
Avoiding Rookie Mistakes in Machine Learning
AI-Powered Ultrasound for Global Maternal Healthcare
DeMo_ Decoupled Momentum Optimization for Large Neural Networks
CS Freshmen and ChatGPT_ A Log Analysis
AI Compiler for Autonomous Vehicles
Competitive Programmer's Handbook
AI Coding Tool Showdown_ Cursor, Bolt, Replit, and V0 Compared
Challenges in Human-Agent Communication
ASL Fingerspelling Recognition Competition
Accelerating Mobile AI with ExecuTorch and KleidiAI