NewsAll

News and Updates

AUG 20, 2026Models

LFM2.5-DSpark: Up to 3.2x Faster Inference from H100 to MacBook

08.20.2026ModelsLFM2.5-DSpark: Up to 3.2x Faster Inference from H100 to MacBook
08.19.2026ModelsLFM2.5 Q4_0: Quantization-Aware Distillation for Edge Deployment
08.18.2026EngineeringDesigning Loops for Production-Grade Work
08.12.2026ModelsLFM2.5-VL-3B: A Better and Faster Vision-Language Model for the Edge
08.05.2026NewsMacPaw Partners with Liquid AI to Bring On-Device AI to Millions of Mac Users
08.04.2026ModelsLFM2.5-2.6B: Deploy Agents Everywhere
07.28.2026ModelsLFM2.5-Encoders: Fast at Long Context, Even on CPU
07.21.2026EngineeringTokenizer Expansion: Upgrading a Model's Tokenizer in Place
07.07.2026EngineeringReducing Doom Loops with Final Token Preference Optimization
06.30.2026EngineeringIFStruct: Measuring structured-output compliance
06.25.2026ModelsLFM2.5-230M: Built to Run Anywhere
06.18.2026ModelsLFM2.5 Retrievers: Bi-directional LFMs for Fast Multilingual Search
06.16.2026PublicationsFixed-Point Reasoners: Stable and Adaptive Deep Looped Transformers
05.31.2026PublicationsNeural Low-Discrepancy Sequences
05.29.2026PublicationsSwitchCraft: A Programmatic Framework for Designing State-Switching Proteins
05.28.2026ModelsLFM2.5-8B-A1B: An Even Better On-Device Mixture of Experts
05.20.2026PublicationsMeMo: Model as a Memory
04.23.2026PressLiquid AI and Mercedes-Benz partner to scale embedded in-car intelligence
04.22.2026PublicationsPreconditioned DeltaNet: Curvature-aware Sequence Modeling for Linear Recurrences
04.08.2026ModelsLFM2.5-VL-450M: Structured Visual Intelligence, Edge to Cloud
03.31.2026ModelsLFM2.5-350M: No Size Left Behind
03.26.2026PublicationsThe Key to State Reduction in Linear Attention: A Rank-based Perspective
03.16.2026External articleWhy Bigger Isn’t Always Better in AI
03.05.2026ModelsNo Cloud, No Waiting: Tool-Calling Agents on Consumer Hardware with LFM2-24B-A2B
03.03.2026PressLiquid AI and Insilico Medicine Announce Strategic Partnership Delivering Lightweight Scientific Foundation Models for Drug Discovery
03.02.2026PublicationsAlphaQ: Calibration-Free Bit Allocation for Mixture-of-Experts Quantization
03.02.2026PublicationsLow-Pass Flow Matching
02.24.2026ModelsLFM2-24B-A2B: Scaling Up the LFM2 Architecture
02.24.2026NewsFrom Cloud to AI PC: Launching Liquid’s Largest LFM2 Model Alongside Our Growing Partner Ecosystem
01.26.2026PublicationsAntislop: A Comprehensive Framework for Identifying and Eliminating Repetitive Patterns in Language Models
01.26.2026PublicationsLearning residue level protein dynamics with multiscale Gaussians