Liquid AI launches LFM3: 2B multimodal models running locally in just 1GB of memory
JosephJacks_ · x · 2026-10-28
Liquid AI officially introduced LFM3, its new generation of Liquid Foundation Models, with a new tri-hybrid architecture and new multimodal encoders. Six model instances across modalities:
- LFM3-1.7B: text, 256k context window
- LFM3-VL-2B: vision-language with video input for spatial understanding
- LFM3-Audio-2B: tool calls, voice chat, reasoning in 18 languages
- LFM3-Omni-2B: omni model combining vision, audio, and language
- d1-2B / d1-Omni-2B: decision models with typed outputs and calibrated probabilities in a single pass; Omni adds audio input
The reposter notes the 2B models handle text, vision, audio and decision-making locally in just 1GB of memory, with the vision model matching Google's Gemma 4 12B on select benchmarks — "small models are getting scary good."
Related event: LiquidAI Releases LFM3 Family of On-Device Models(15 posts)→
More from Models
- Surge AI launches DAYJOB: Tax benchmark — Opus caught the deadline 5/5 times, Astra never did — echen · 2026-10-28
- ~65% of TAU-banking agent benchmark tasks found defective — HamelHusain · 2026-10-28
- Opus 5.5 dominant almost everywhere, but GPT 6 Astra still wins at writing — brandon_galang · 2026-10-28
- Report: OpenAI's unreleased internal model Astra billed as start of the AGI era, internal AGI possible by year-end — haider1 · 2026-10-28
- ChatGPT Plus users report memory loss, slow responses, and sudden personality change — International-Top951 · 2026-10-28
- European 'sovereign AI' startup Giotto bans benchmarking in its ToS, drawing flak — teortaxesTex · 2026-10-28