Nano-Refuse-0.4BvsRetrace-1.5B
Decision Card metrics — latency, license, lineage overlap, benchmarks, and repo health from the live graph.
Model ANano
Nano-Refuse-0.4B
1graph score
Open model page →Model BRetraceRetrace-1.5B
1graph score
Open model page →Nano-Refuse-0.4B leads on graph popularity (1 vs 1)
Decision metrics
| Attribute | Nano-Refuse-0.4B | Retrace-1.5B |
|---|---|---|
| License | — | — |
| ParametersModel size | — | — |
| Context window | — | — |
| Architecture | — | — |
| Hardware class | — | — |
| Latency | — | — |
| $ / 1M tokens | — | — |
| Verified TTFT | — | — |
| Repo health (RIS) | — | 24 |
| Graph trust | 88 | 88 |
| Lineage | No lineage links | 10 papers · 1 repo |
| Graph score | 1 | 1 |
Decision cards
Nano-Refuse-0.4B
License—
Parameters—
Context—
Architecture—
Lifecycle
Lineage
BenchmarksPending verified runs
Repo healthNo linked repo
Economics—Economics pending verified runs
Retrace-1.5B
License—
Parameters—
Context—
Architecture—
Lifecycle
Lineage
BenchmarksPending verified runs
Repo health
RIS 24estimated
retrace-agents →Economics—Economics pending verified runs
Retrace-1.5B
A small reasoning model tuned to self-correct via failure traces.