LFM2.5-1.2B Instruct: benchmark results
Provider: Liquid AI. Released 2025-06-17. Access: Open.
Unified ELO 1359 ± 1, rank #1250 of 1392 rated models, from 322 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| MERA Code - CodeCorrectness | 84.5 | Accuracy (%) | 89.2 |
| AI Chess Leaderboard (Reasoning) | 755 | Elo | 61.3 |
| LIBRA - ruQuALITY | 71.4 | Dataset Total Score (%) | 50 |
| AA IFBench | 43.81 | Accuracy (%) | 48.6 |
| MERA Code - CodeLinterEval | 48.27 | pass@1 (%) | 47.7 |
| EuroEval Catalan NLU - ScaLA CA | 2.5 | Linguistic acceptability Score (%) | 44.5 |
| AA Humanity's Last Exam | 6.72 | Accuracy (%) | 43.8 |
| EuroEval Portuguese NLU - HAREM | 36.05 | Named entity recognition Score (%) | 34.6 |
| EuroEval Spanish NLU - Sentiment Headlines ES | 32.04 | Sentiment classification Score (%) | 34.2 |
| EuroEval Dutch Simplification - Duidelijke Taal | 44.51 | Score (%) | 34.1 |
| LIBRA - ruTPO | 70.4 | Dataset Total Score (%) | 33.3 |
| EuroEval Portuguese NLU - ScaLA PT | 3.91 | Linguistic acceptability Score (%) | 32.6 |
Interactive version: theaggregate.ai/model?slug=lfm2-5-1-2b-instruct · How It Works · Data refreshed daily, snapshot 2026-09-05.