Liquid AI launches LFM3: tri-hybrid architecture, six edge models, all modalities
maximelabonne · x · 2026-10-28
Liquid AI introduced LFM3, its new generation of Liquid Foundation Models, built on a new tri-hybrid architecture with new multimodal encoders, targeting specialized intelligence on edge devices. Six model instances:
- LFM3-1.7B: text model with a 256k context window
- LFM3-VL-2B: vision-language model with video-input support 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
Maxime Labonne (Liquid AI) calls the 1.7B an "absolute (agentic) monster".
Related event: Liquid AI releases LFM3: six on-device models spanning all modalities(4 posts)→
More from Models
- mark_k pans rumored OpenAI Terra and Astra models, calls for a stronger Sol tier — mark_k · 2026-10-28
- Kimi K3.1 teased to adopt DSA like DeepSeek V3.2 for better inference economics — zephyr_z9 · 2026-10-28
- LiquidAI releases LFM3 weights on Hugging Face: seven models from 1.7B text to 2B omni — omarsar0 · 2026-10-28
- Cisco launches Model Foundry: partners can deploy and fine-tune its purpose-built domain AI models — aminkarbasi · 2026-10-28
- wyat-2: first decision model built on a looped transformer, trained on a single RTX 4070 Ti — Fit_Strawberry8480 · 2026-10-28
- OpenAI opens Decisions API beta: up to 10x faster decisions than GPT-6 Luna — OpenAIDevs · 2026-10-28