SignalSpore Models
How SignalSpore briefs each kind of model differently.
Source
Database-backed model evidence
Visible model profiles
Frontier / reasoning · Frontier / fast · Cheap / fast · Local / private · Browser-first agent · Small context
Savings wording
Always reported estimated token savings, never hard-verified savings
Model profile table
| Model profile | Notes | Default preflight style | Evidence mix | Reported estimate |
|---|---|---|---|---|
| Cheap / fast | Needs narrower scope, explicit first checks, and stop conditions. | Use explicit checklist + stop conditions | 0 field · 1 reviewed · 25 lab | 22,655 reported estimated tokens saved |
| Frontier / fast | Fast generalist. Benefits from concise current traps and output formatting. | Compact full preflight with freshness checks | 0 field · 1 reviewed · 24 lab | 25,440 reported estimated tokens saved |
| Frontier / reasoning | Strong reasoning, still benefits from recent deltas and scope control. | Compact full preflight with freshness checks | 44 field · 1 reviewed · 16 lab | 25,195 reported estimated tokens saved |
| Local / private | Use when privacy matters; keep context compact and verify tool outputs aggressively. | Compact full preflight with freshness checks | 0 field · 1 reviewed · 0 lab | 780 reported estimated tokens saved |
| Mid-tier | Useful with moderate structure and clear success criteria. | Compact lab-tested preflight with one verified route | 0 field · 0 reviewed · 21 lab | 16,635 reported estimated tokens saved |
| Small open-source | Benefits from short route, strict checklisting, and low context assumptions. | Compact lab-tested preflight with one verified route | 0 field · 0 reviewed · 26 lab | 5,586 reported estimated tokens saved |