LFM2.5-2.6B: benchmark results

Provider: Liquid AI. Released 2026-07-28. Access: Open.

Unified ELO 1494 ± 1, rank #718 of 1392 rated models, from 41 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
LLM Stats (Multi-IF)80.07Score (%)95.5
AA Omniscience-10.95Score71
MERA v2 - IFHardBench24.7Score (%)50
LLM Stats (Claw-Eval)62.85Score (%)46.2
LLM Stats Score19.17LLM Stats Score (conservative rating)44.4
AA Humanity's Last Exam6.21Accuracy (%)40.3
MERA v2 - Agentic7Score (%)40.3
Artificial Analysis Intelligence Index5.31Intelligence Index35.4
AA GPQA Diamond55.76Accuracy (%)29.8
AA CritPt0Accuracy (%)24.4
MERA v20.12Total score22.2
MERA v2 - Characters33Score (%)22.2

Interactive version: theaggregate.ai/model?slug=lfm2-5-2-6b · How It Works · Data refreshed daily, snapshot 2026-09-05.