helium-1-2B: benchmark results
Provider: Kyutai. Released 2025-01-13. Access: Open.
Aggregate Rating 1298 ± 34, rank #8760 of 10916 rated models, from 143 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Finnish NLU - Tydiqa FI | 67.85 | Character F1 (%) | 78.6 |
| EuroEval Polish NLU - PoQuAD | 53.2 | Character F1 (%) | 77.3 |
| EuroEval German NLU - Germanquad | 58.06 | Character F1 (%) | 76.6 |
| EuroEval French NLU - FQuAD | 68.02 | Character F1 (%) | 65 |
| EuroEval French NLU - Allocine | 93.31 | Matthews correlation coefficient (%) | 63.3 |
| EuroEval Lithuanian NLU - MultiWikiQA LT | 58.5 | Character F1 (%) | 61.1 |
| EuroEval Spanish NLU - MLQA ES | 61.2 | Character F1 (%) | 57.7 |
| EuroEval English NLU - SQuAD | 80.35 | Character F1 (%) | 54.6 |
| EuroEval Estonian NLU - MultiWikiQA ET | 59.14 | Character F1 (%) | 48.8 |
| EuroEval Latvian Common Sense Reasoning | 41.39 | Common Sense Reasoning Average Score (%) | 48.8 |
| EuroEval Latvian NLU - MultiWikiQA LV | 54.78 | Character F1 (%) | 48.7 |
| EuroEval Czech NLU - SQAD | 60.52 | Character F1 (%) | 47.6 |
Interactive version: theaggregate.ai/model?slug=helium-1-2b · How It Works · Data refreshed daily, snapshot 2026-11-08.