A 14-minute video traces one ChatGPT question from the Enter key to the GPU and back
Glass-Garden-5888 · reddit · 2026-11-13
A 14-minute video follows one question from pressing Enter to the GPU and back: the message leaves as JSON, crosses edge servers and 6,400 km of transatlantic cable; OpenAI's front door handles login, rate limits and tokenization; in the data centre a rack of 72 GPUs batches your query with thousands of others; inside the model, embeddings, a real attention head, MLP neurons, 12 layers, the softmax vote, temperature, and the loop that emits one token at a time; the answer then streams back word by word. Since OpenAI's internals aren't public, the model segment uses open-weights GPT-2 small (same design) with every on-screen number measured; illustrative ChatGPT-side parts are explicitly labeled. Narration is an AI voice.
More from Infra
- Analyst argues labs already run Blackwell stacks, with 3-6x smuggling premium — teortaxesTex · 2026-11-13
- Chinese major models reportedly trained on Blackwell-class chips, though scale disappoints — teortaxesTex · 2026-11-13
- Sentdex: 'China spy' claim is likely BS — more likely mutual export-restriction circumvention — Sentdex · 2026-11-13
- Analyst: DeepSeek V4's Flash variant is winning, and a hyper-efficient 2B V5 could be the real play — teortaxesTex · 2026-11-13
- vLLM ships recipe for Thinking Machines' Inkling-2-Small: 594B MoE, 1M context, ~1,100 tok/s/GPU — vllm_project · 2026-11-13
- Inkling 2 Small serving details: ~1,070 TPGS agentic throughput at 32K context — vllm_project · 2026-11-13