Lucie-7B: benchmark results
LINAGORA and OpenLLM-France's Lucie-7B, a pretrained base model trained on 3 trillion tokens split across English, French, German, Spanish, Italian and programming languages. Provider: OpenLLM-France. Released 2024-10-10. Access: Open.
Aggregate Rating 1358 ± 30, rank #1365 of 1656 rated models, from 52 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard v1 - TruthfulQA MC2 | 67.67 | MC2 (%) (0-shot) | 86.4 |
| Open LLM Leaderboard v1 - HellaSwag | 84.81 | Normalized accuracy (%) (10-shot) | 68.9 |
| Open LLM Leaderboard v1 - ARC Challenge | 62.2 | Normalized accuracy (%) (25-shot) | 56.6 |
| EuroEval Spanish NLU - Sentiment Headlines ES | 41.13 | Matthews correlation coefficient (%) | 55.6 |
| Open LLM Leaderboard v1 - GSM8K | 39.58 | Accuracy (%) (5-shot) | 54.6 |
| Open LLM Leaderboard v1 - WinoGrande | 76.8 | Accuracy (%) (5-shot) | 50.7 |
| Open LLM Leaderboard v1 - MMLU | 60.34 | Accuracy (%) (5-shot) | 47.8 |
| EuroEval Italian NLU - Sentipolc16 | 46.33 | Matthews correlation coefficient (%) | 43 |
| CARTE-LV | 44.2 | Accuracy (%) on the 233 CARTE-LV questions on regional lingu | 42.3 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 65.42 | Character F1 (%) | 39.2 |
| Open LLM Leaderboard - MuSR | 39.23 | Score | 36.6 |
| CARTE - Corse | 51 | Accuracy (%) on the 100 CARTE questions about Corse (Corsica | 36.5 |
Interactive version: theaggregate.ai/model?slug=lucie-7b · How It Works · Data refreshed daily, snapshot 2026-10-20.