Liquid AI's Open-Source 1.7B Model Tops Multi-IF, Beating ERNIE 5.0

Liquid AI's open-source LFM3-1.7B achieved SOTA on Meta's Multi-IF benchmark for multi-turn, multilingual instruction following, scoring 89.1 and surpassing Baidu's much larger closed-source ERNIE 5.0.

2026-11-11 ~ 2026-11-12 · 2 related posts

• LiquidAI's 1.7B LFM3 hits SOTA on Multi-IF, beating 100x-larger rivals including ERNIE 5.0 — NielsRogge · 2026-11-11
• Liquid AI's 1.7B open model beats Baidu's ERNIE 5.0 on multi-turn instruction following — JosephJacks_ · 2026-11-12