9 guides
Guides
The reasoning behind the numbers in the TokenPad tools: what a token is, why the same text costs different amounts on different models, and where an API bill actually goes. Each one ends somewhere you can act on it.
- Start here7 minWhat is a token in an LLM?The unit everything else is denominated in. Start here if tokens are still fuzzy.August 3, 2026
- Start here11 minHow to reduce LLM API costsNine levers, ordered by what they return per hour of work. Most teams find 40% in the first three.August 3, 2026
- 8 minPrompt caching, and what it actually savesThe single largest lever on a repetitive workload — and the cases where it does nothing.August 3, 2026
- 6 minContext window vs token limit vs max tokensThree terms, three different limits, one very common production bug.August 3, 2026
- 7 minWhy your JSON costs three times more tokens than you thinkHalf a JSON payload can be structure the model gains nothing from. Here is how to find it.August 3, 2026
- 9 minChunking strategies for RAG that actually retrieveChunk size, overlap and boundaries — and how to tell which one is breaking your retrieval.August 3, 2026
- 8 minWhy chatbot costs grow faster than your user countTurn 20 pays for turns 1 to 19 again. Why per-request budgets are wrong by 3–4×.August 3, 2026
- 9 minWhat an AI agent actually costs to runTool schemas and iteration limits multiply. Where agent bills come from, itemised.August 3, 2026
- 8 minLLM API pricing compared, and how to read itThe headline rate is the least useful number on the page. What to compare instead.August 3, 2026
Where to start
If tokens are still fuzzy, read what a token actually is first — everything else assumes it. If you are here because a bill surprised you, go straight to how to reduce LLM API costs, which is ordered by return on effort rather than by how clever each idea is.
Every guide routes into the tools, and every number quoted traces back to the model price table, where each figure carries the provider URL it came from and the date it was read.