Module 2.3 · Topic 2
How Different Models Process Inputs
Bottom Line Up Front: Identical prompts produce different outputs across model tiers because each model processes information differently. A fast model generates a response directly; a reasoning model deliberates first;…
2.1 Why the Same Prompt Yields Different Results Across Models
Diagnose output quality by holding the prompt constant and varying the model tier. If all models struggle, the prompt needs work. If only the fast model struggles, you need a reasoning model.
Logic behind this approach:
A weak fast-model response may reflect model capacity, not prompt quality. Testing identical prompts across tiers isolates whether the gap is prompt-related or model-related.
Sample prompt:
What to expect in reply:
A fast model delivers a matrix quickly but may miss trade-offs like hidden verification costs. A reasoning model shows its deliberation, surfaces second-order effects, and explains its weighting.
2.2 How Reasoning Models Use Extended Thinking
Reasoning models allocate computation time to internal deliberation (extended thinking) before responding. A fast model generates output directly; a reasoning model pauses to think. You see the reasoning unfold, showing deliberation and reconsideration of assumptions. This can take 10-60 seconds or longer but produces evidence you can verify. If reasoning contains a flaw, you catch it before acting. For high-stakes decisions, this transparency is essential. For routine tasks, reasoning models are overkill.
Unlock the Full AI Skill Building Experience
Visit LawQi for access to the full AI Skill Building experience.
Visit LawQi40% discount using code REASONABLE for personal subscriptions.
Free 48-hour preview access when investigating for teams.