calme-3.3-instruct-3B — benchmark results
MaziyarPanahi's general-purpose fine-tune of Qwen2.5-3B, trained on arcee-ai EvolKit-20k and his French instruct dataset. Provider: Other. Released 2024-11-07. Access: Open.
Unified ELO 1464 ± 38, rank #939 of 1776 rated models, from 10 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - MATH Level 5 | 37.39 | Score | 87.8 |
| Open LLM Leaderboard - IFEval | 64.23 | Score | 76.6 |
| French LLM Leaderboard - IFEval FR | 49.67 | Score (%) | 68.5 |
| French LLM Leaderboard - Average | 35.59 | Average Score (%) | 64.8 |
| French LLM Leaderboard - BAC FR | 30.37 | Score (%) | 61.1 |
| Open LLM Leaderboard - MMLU-Pro | 25.62 | Score | 46.3 |
| Open LLM Leaderboard - MuSR | 9.4 | Score | 45.6 |
| Open LLM Leaderboard - BBH | 25.68 | Score | 39.5 |
| Open LLM Leaderboard - GPQA | 4.36 | Score | 37.6 |
| French LLM Leaderboard - GPQA FR | 26.74 | Score (%) | 28.7 |
Interactive version: theaggregate.ai/model?slug=calme-3-3-instruct-3b · How the rankings work · Data refreshed daily, snapshot 2026-07-22.