Tauseef FayyazA to Z
Glossary
Every concept in this sheet, plus the smaller terms it uses. Select a concept to open its card.
B
- Benchmarks
- Fixed public test sets with a published score per model, for coding, maths, knowledge and agent tasks.
C
- Compounding errors
- In a task with many steps, per-step accuracy multiplies: 95% per step over 10 steps is about 60% overall.
D
- Dated snapshots
- Every table of model names, prices or context windows is a photograph of one day.
E
- Effort and thinking settings
- A request setting that controls how much a model reasons: Anthropic calls it effort, OpenAI reasoning effort, Gemini a thinking setting.
I
- Instruction following
- How faithfully a model obeys your format, length, language, tone and must-not rules, including many turns into a conversation.
L
- Leaderboards
- Public rankings of models.
M
- Model requirements card
- Six lines written before any testing: quality (and the cost of a wrong answer), speed, volume, context size, tools and privacy.
- Model tiers
- Providers sell a family in sizes: a top tier (most capable, slowest), a middle workhorse and a small, cheap, fast tier.
P
- Price plus cost of errors
- Total cost per request = model price + (error rate x cost of one wrong answer).
R
- Reasoning model
- A model that writes hidden working before its final answer.
- Reasoning tokens
- The tokens a model spends thinking before its visible answer.
T
- Task families
- Five common kinds of LLM work: chat, coding, research, data analysis and workflow automation.
- The cheapest model that is good enough
- The model to ship is the cheapest and fastest one that clears your quality bar for this task, not the smartest one available.
W
- Which model for which job
- A starting point for each kind of job: which tier to test first and when to step up or down.