1.2.1 · Lesson 4 of 28
Assess and choose FMs for business use cases
Select a model based on:
WHAT
Select a model based on:
- capability
- quality
- latency
- cost
- context window
- modality
- reasoning capability
- structured output capability
- availability
- regional constraints
WHY
Different FMs have different strengths.
The correct model depends on the workload.
HOW
Use a model-selection matrix:
| Requirement | Model characteristic |
|---|---|
| Complex reasoning | Strong reasoning capability |
| High-volume simple tasks | Lower-cost / lower-latency model |
| Image understanding | Multimodal capability |
| Long documents | Appropriate context capability |
| Structured output | Strong structured-output support |
| Sensitive deployment | Appropriate security / deployment model |
Evaluate using representative workloads rather than assumptions.
AWS services
- Amazon Bedrock
- Amazon SageMaker AI
- Bedrock model evaluation capabilities
WHEN TO USE
When several FMs can solve the same problem.
Common mistake
Choosing a model solely because it has the highest benchmark score.
EXAM TRAP
Benchmark score alone does not determine the best model.
Business-specific evaluation matters.
---
Key points
- Different FMs have different strengths.
- The correct model depends on the workload.
- When several FMs can solve the same problem.
- Choosing a model solely because it has the highest benchmark score.
Exam focus
Benchmark score alone does not determine the best model. Business-specific evaluation matters.