Liquid AI Expands LFM2 Tokenizer to 128K for Multilingual Efficiency

Liquid AI expanded the LFM2 tokenizer vocabulary from 65K to 128K without retraining from scratch. This upgrade significantly improves multilingual efficiency, reducing Thai tokens by four times and boosting on-device decoding speeds by up to 3.7 times.

2026-07-21 ~ 2026-07-22 · 3 related posts

• LFM2 tokenizer expansion cuts Thai tokens 4× and speeds on-device decoding up to 3.7× — maximelabonne · 2026-07-21
• Liquid AI expands a pretrained tokenizer from 65K to 128K without retraining from scratch — JosephJacks_ · 2026-07-21
• LFM2.5-8B-A1B doubles its tokenizer vocab and cuts on-device decoding time up to 3.7x — maximelabonne · 2026-07-22