CARTE-LV: leaderboard
CARTE-LV (2026) is the 233-question linguistic-variation subset of the French CARTE benchmark, testing knowledge of regional French grammar, discourse markers, pragmatics, register and lexical usage in 13 metropolitan regions. It is scored as 0-shot multiple-choice accuracy in percent with an I do not know option; higher is better.
Metric: Accuracy (%) on the 233 CARTE-LV questions on regional linguistic variation (grammar, discourse markers, pragmatics, register and region-specific lexical usage in 13 regions), 0-shot, five options including 'Je ne sais pas'; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 93.6 | #164 |
| 2 | Mistral Nemo Instruct (2407) | 74.2 | #2947 |
| 3 | Gemma 3 12B (IT) | 68.7 | #1837 |
| 4 | Qwen 3.5 9B | 65.7 | #803 |
| 5 | EuroLLM-9B-Instruct | 65.2 | #3589 |
| 6 | Llama 3.1 8B Instruct | 63.9 | #4246 |
| 7 | Qwen 3.5 4B | 58.4 | #1175 |
| 8 | aya-expanse-8B | 58.4 | #3885 |
| 9 | Mistral 7B Instruct (v0.2) | 57.9 | #7007 |
| 10 | Mistral 7B Instruct (v0.3) | 57.5 | #6653 |
| 11 | Luth-LFM2-1.2B | 49.4 | #7549 |
| 12 | Llama 3.2 3B Instruct | 48.9 | #7147 |
| 13 | Lucie-7B | 44.2 | #8468 |
| 14 | Llama 3.2 1B Instruct | 40.8 | #9047 |
| 15 | gemma-4-E4B-it | 23.2 | #1469 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=carte-lv · How It Works · Data refreshed daily, snapshot 2026-11-13.