The trade-offs, stated honestly.

An agent lets the model decide what steps to take and in what order; a workflow is a fixed, developer-defined sequence the model fills in one step at a time.

Closed models are served only through a provider's API with weights never released; open-weight models publish their weights for anyone to download, run, and fine-tune.

A code grader scores AI output with deterministic rules — exact match, regex, tests passing. An LLM judge scores it with another model reasoning against a rubric.

Zero-shot asks a model to do a task from instructions alone; few-shot adds a handful of worked examples so the model can pattern-match instead of just following a description.

Both shrink a model's weights from 16-bit floats to lower-precision integers to cut memory and speed inference — INT8 keeps more precision, INT4 keeps less and shrinks the model further.

Prompt engineering shapes model behavior through instructions and examples at request time; fine-tuning changes the model's weights through additional training.

RAG retrieves relevant documents at query time and grounds the answer in them; fine-tuning further trains the model's weights on your own examples.

Chunk size is one of the cheapest, highest-leverage dials in a RAG pipeline — small chunks retrieve precisely but lose context; large chunks preserve context but retrieve fuzzily.

Temperature reshapes how sharply a model favors its top-probability tokens; top-p trims which tokens are even eligible to be sampled. They tune together, not against each other.

Vector search ranks results by embedding similarity, matching meaning even without shared words; keyword search ranks by exact term overlap.