LFM2 2.6B: benchmark results

Provider: Liquid AI. Released 2025-11-28. Access: Open.

Unified ELO 1386 ± 1, rank #1185 of 1392 rated models, from 333 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Spanish NLU - Sentiment Headlines ES46.63Sentiment classification Score (%)75.7
MERA - LCS16.2Accuracy (%)66
MERA Code - CodeCorrectness75.39Accuracy (%)63.1
MERA Code - stRuCom25.91chrF (%)63.1
EuroEval Portuguese NLU - MultiWikiQA PT70.53Reading comprehension Score (%)60.6
EuroEval Spanish44.67Average Score (%)59
EuroEval Portuguese Knowledge45.69Knowledge Average Score (%)56.6
EuroEval Spanish Common Sense Reasoning46.43Common Sense Reasoning Average Score (%)56.5
EuroEval Spanish NLU44.13NLU Average Score (%)56.1
EuroEval Italian Common Sense Reasoning42.83Common Sense Reasoning Average Score (%)55.7
EuroEval Portuguese47.02Average Score (%)55.3
EuroEval Italian NLU - Sentipolc1651.5Sentiment classification Score (%)55.2

Interactive version: theaggregate.ai/model?slug=lfm2-2-6b · How It Works · Data refreshed daily, snapshot 2026-09-05.