Task
Use-case capability comes first.
Methodology
The Stackelier Method
We do not rank what is popular. We rank what fits the task, privacy boundary and hardware — then show the evidence.
Privacy
Application, storage and inference location are separate.
Hardware
Model + runtime + quantization + context define fit.
Evidence
Critical claims must be traceable.
Ranking weights
| Use-case | 30% |
|---|---|
| Privacy | 25% |
| Hardware | 20% |
| Professional fit | 10% |
| Complexity | 5% |
| Cost | 5% |
| Evidence confidence | 5% |