Interactive learning for modern LLMs

Understand LLMs by feeling them

Explore latency, tokenization, retrieval (RAG), network structure, sampling, and agents—hands-on, in your browser. No accounts. No backend. Just insight.

Runs 100% in-browserNo sign-in requiredOptimised for WebGPU
Latency shapes UX
  • Vary tokens/sec to feel cadence
  • Perceived speed vs. actual throughput
  • Design for anticipation
Tokens are the atoms
  • Whitespace vs. model tokens
  • Why prompts ≠ characters
  • Costs scale with tokens
RAG = open-book answers
  • Retrieve relevant snippets
  • Fit within a context window
  • Answer with citations

Ready to peek under the hood?

Jump into any demo—each one is fast, focused, and teaches by doing.