Liquid AI launches LFM3 model family; 1.7B hits 185 tok/s on-device
JosephJacks_ · x · 2026-10-30
Liquid AI officially announced LFM3, a new generation of foundation models with a tri-hybrid architecture and multimodal encoders, spanning six instances across text, audio, vision and omni modalities, plus decision models (d1) with typed outputs and calibrated probabilities in a single pass. Highlights: LFM3-1.7B (text, 256k context), LFM3-VL-2B (video input for spatial understanding), LFM3-Audio-2B (tool calls, voice chat, 18 languages), and LFM3-Omni-2B. A reposter shared on-device benchmarks of LFM3-1.7B (Q80) on an M3 Ultra Mac via LM Studio: 1.85GB, 185 tokens/s generation, 0.08s time-to-first-token.
More from Models
- Researchers Observe Opus 3's Simulated Characters Almost Always Share Its Own Values — repligate · 2026-10-30
- evident-charts: open-source skill teaches coding agents to make honest charts and self-check — randal_olson · 2026-10-30
- Elorian launches visual reasoning model Glimpse alongside DecoyCount benchmark — AndrewDai · 2026-10-30
- Instruction Capacity paper quantifies rule-following limits: GPT-4o mini holds ~37 rules at 95% accuracy — Hennadii Kutomanov · 2026-10-30
- Guidelabs launches Causal Diffusion Engine, adding interpretability to diffusion LLMs with no latency cost — juliusadml · 2026-10-30
- DeepSeek's cyber model enters internal testing; observers back keeping it closed — Xianbao_QIAN · 2026-10-30