Trinity-Nano-Base: benchmark results
Arcee AI Trinity Nano base checkpoint before fine-tuning: 56 layers, two dense and 54 MoE layers, routing to eight of 128 experts plus one shared expert. Its publisher reports 6B total, 1B active parameters and 10T training tokens. Provider: Arcee AI. Released 2025-12-01. Access: Open.
Aggregate Rating 1281 ± 29, rank #8971 of 10916 rated models, from 240 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Dutch Simplification - Duidelijke Taal | 50.3 | Score (%) | 59.7 |
| EuroEval English Knowledge | 69.28 | Knowledge Average Score (%) | 44.3 |
| EuroEval Catalan NLU - ScaLA CA | 1.31 | Matthews correlation coefficient (%) | 42.7 |
| EuroEval Italian NLU - ScaLA IT | 12.93 | Matthews correlation coefficient (%) | 40.5 |
| EuroEval Latvian Common Sense Reasoning | 17.78 | Common Sense Reasoning Average Score (%) | 37.9 |
| EuroEval Ukrainian NLU - NER UK | 41.49 | Micro F1 excluding MISC (%) | 36 |
| EuroEval Icelandic NLU - MIM-GOLD NER | 37.73 | Micro F1 excluding MISC (%) | 35.8 |
| EuroEval Slovene Common Sense Reasoning | 4.08 | Common Sense Reasoning Average Score (%) | 35.5 |
| EuroEval Norwegian Knowledge | 8.21 | Knowledge Average Score (%) | 35.4 |
| EuroEval Icelandic Knowledge | 4.36 | Knowledge Average Score (%) | 35.2 |
| EuroEval Czech NLU - PONER | 38.55 | Micro F1 excluding MISC (%) | 35.1 |
| EuroEval Norwegian Knowledge - NRK Quiz QA | 16.25 | MCC (x100) | 35 |
Interactive version: theaggregate.ai/model?slug=trinity-nano-base · How It Works · Data refreshed daily, snapshot 2026-11-08.