Sabia-4: benchmark results
Provider: Other.
Aggregate Rating 1682 ± 45, rank #438 of 10916 rated models, from 6 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MARCA (English, Orchestrator) | 83.2 | Checklist accuracy (0-1, times 100): share of a question's c | 84.6 |
| P3B3 (pt-BR Prompt) | 4.6 | Variety score (0-100), where 0 is Brazilian and 100 European | 84.2 |
| MARCA (Portuguese, Orchestrator) | 80 | Checklist accuracy (0-1, times 100): share of a question's c | 61.5 |
| BLUEX v2 (Official) | 8.6 | Mean rubric score (0-10; 880 subquestions; evaluation errors | 60 |
| MARCA (English, Basic) | 74.3 | Checklist accuracy (0-1, times 100): share of a question's c | 53.8 |
| MARCA (Portuguese, Basic) | 69.5 | Checklist accuracy (0-1, times 100): share of a question's c | 46.2 |
Interactive version: theaggregate.ai/model?slug=sabia-4 · How It Works · Data refreshed daily, snapshot 2026-11-08.