Trinity-Nano-Preview: benchmark results
Arcee AI's experimental Trinity Nano preview, a 6B-total/1B-active text MoE chat model. Provider: Arcee AI. Released 2025-12-01. Access: Open.
Aggregate Rating 1365 ± 22, rank #7287 of 10916 rated models, from 268 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Portuguese NLU - ScaLA PT | 16.75 | Matthews correlation coefficient (%) | 60 |
| EuroEval Spanish NLU - ScaLA ES | 20.05 | Matthews correlation coefficient (%) | 57.9 |
| EuroEval Catalan NLU - ScaLA CA | 12.14 | Matthews correlation coefficient (%) | 57.8 |
| EuroEval Catalan NLU - WikiANN CA | 64.15 | Micro F1 excluding MISC (%) | 54 |
| EuroEval English Knowledge | 75.33 | Knowledge Average Score (%) | 51.8 |
| EuroEval Portuguese NLU - HAREM | 41.87 | Micro F1 excluding MISC (%) | 51.8 |
| EuroEval Bulgarian NLU - MultiWikiQA BG | 62.06 | Character F1 (%) | 49.8 |
| EuroEval Icelandic NLU - ScaLA IS | 3.72 | Matthews correlation coefficient (%) | 49.2 |
| EuroEval Icelandic NLU - MIM-GOLD NER | 44.6 | Micro F1 excluding MISC (%) | 47.3 |
| EuroEval English NLU - ScaLA EN | 41.9 | Matthews correlation coefficient (%) | 46.2 |
| EuroEval Estonian NLU - EstNER | 42.21 | Micro F1 excluding MISC (%) | 46.2 |
| Medmarks - MedHallu Hard | 44.43 | Score (%) | 45.5 |
Interactive version: theaggregate.ai/model?slug=trinity-nano-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.