How GPT, Claude, and Gemini are actually trained and served – Reiner Pope

Explains the batching and memory tradeoffs behind latency tiers and API prices, so you can predict cost and speed constraints when designing agent workloads.

• token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
• AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.

• videoTerence Tao – How the world’s top mathematician uses AI
• videoJensen Huang – Will Nvidia’s moat persist?
• videoThe OpenAI/Hugging Face attack, clearly explained
• videoAndrej Karpathy — “We’re summoning ghosts, not building animals”

• podcastReiner Pope – The math behind how LLMs are trained and servedDwarkesh Patel
• videoVertical Mobility: Inference from MVP to Trillion-Parameter Workloads — Sitanshu Gupta, CoreWeaveAI Engineer
• videoGroq Founder, Jonathan Ross: OpenAI & Anthropic Will Build Their Own Chips & Will NVIDIA Hit $10TRN20VC

Checking sign-in…

Loading comments…