Computerphile explains why LLM tokens are expensive

Category: Video
Status: discussion
Sources: watch
Channel: Computerphile (2026-07-02, 201k views, 5.0 over 8980 ratings)
Summary: A Computerphile explainer walks through why large-language-model tokens carry the cost they do, connecting tokenization, context length, and per-token compute to the pricing developers see. It is an educational overview rather than a product announcement.
Why it matters: The token-economics framing is the backdrop to practical cost decisions in agentic and API workloads, including the day's image-token arbitrage discussion.