Sabiá-3.1: benchmark results
Provider: Other.
Aggregate Rating 1575 ± 50, rank #1328 of 10916 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Magis-Bench | 4.41 | AVG (self-reported) | 63.6 |
| Prosa | 60.5 | Score (self-reported) | 26.7 |
| CAPITU - Coherence | 93.7 | Coherence score (%): a GPT-4o-mini judge's 0-1 rating of the | 17.6 |
| CAPITU - Easy | 90 | Strict prompt-level accuracy (%) on the 70 easy prompts (one | 17.6 |
| CAPITU - Multi-Turn | 64 | Conversation accuracy (%): share of the 100 three-turn CAPIT | 17.6 |
| CAPITU - Hard | 53.3 | Strict prompt-level accuracy (%) on the 60 hard prompts (thr | 5.9 |
| CAPITU - Instruction-Level Accuracy | 85.4 | Instruction-level accuracy (%): share of all individual cons | 5.9 |
| CAPITU - Strict Accuracy | 74 | Strict prompt-level accuracy (%): share of prompts whose con | 5.9 |
| CAPITU | 89.2 | Final score (%): the mean over prompts of (2 x the share of | 2.9 |
| CAPITU - Medium | 75.7 | Strict prompt-level accuracy (%) on the 70 medium prompts (t | 0 |
Interactive version: theaggregate.ai/model?slug=sabi-3-1 · How It Works · Data refreshed daily, snapshot 2026-11-08.