Architecture decisions
LLM comparison guides
Use these guides when two related techniques solve different parts of the problem. Each comparison provides a decision table, example, and hands-on exercise.
Tokens vs Embeddings
Tokens are discrete text units; embeddings are learned numeric representations used to compare or process meaning.
Open guideRAG vs Fine-Tuning
RAG supplies external evidence at request time; fine-tuning changes model behavior by updating parameters or adapters.
Open guideTool Calling vs MCP
Tool calling is the model-facing action pattern; MCP standardizes how clients discover and communicate with capability providers.
Open guideAgents vs Workflows
Workflows follow explicit control logic; agents delegate some next-action selection to a model within bounded state and policy.
Open guidePrompting vs Structured Outputs
Prompting communicates intent in language; structured outputs add a machine-checkable response contract.
Open guideLong Context vs Retrieval
Long context supplies more text directly; retrieval selects a smaller evidence set for each request.
Open guideLoRA vs Full Fine-Tuning
LoRA trains small low-rank adapters while freezing the base model; full fine-tuning updates the full parameter set.
Open guideQuantization vs Distillation
Quantization uses lower-precision numbers; distillation trains a smaller student to reproduce useful teacher behavior.
Open guide