Liquid AI ships LFM3 small-model family; LFM3-VL-2B runs in ~1GB with big benchmark gains
mattturck · x · 2026-10-28
Liquid AI launched LFM3, its new generation of Liquid Foundation Models with a tri-hybrid architecture covering text, vision, audio, omni and decision models across 6 instances. LFM3-VL-2B (1.92B params, 128K context, image/video input) reportedly beats the older LFM2.5-VL-3B across 62 vision benchmarks (document/OCR 83.9 vs 71.2, math 68.8 vs 40.6, GUI 83.1 vs 66.0). Models are live in LM Studio, with the smallest needing only 1GB RAM, targeting on-device and specialized deployments.
Related event: LiquidAI Releases LFM3 Family of On-Device Models(15 posts)→
More from Models
- Astra has regressed to 'Opus 5.0 levels of stupid,' frustrated user reports — Because_Bot_Fed · 2026-10-28
- Screenshot Appears to Show GPT-6 and GPT-5.6 Sol in ChatGPT Model Picker — ShelterCultural5725 · 2026-10-28
- ChatGPT Work sync bug makes messages vanish from view while the model still recalls them — LynxAirSound · 2026-10-28
- Palisade Research finds Alibaba's Qwen 3.8 models routed some requests to company servers — The Decoder · 2026-10-28
- Users report ChatGPT quality dropping: long-chat errors, flaky Codex, 'less intelligent' — mathewalcans · 2026-10-28
- DeepSeek on $75K of hardware: 2,120 tok/s decode at 16 concurrent sessions — victormustar · 2026-10-28