Liquid AI

Liquid AI launched Liquid RIPL, an inference engine that runs LFMs in under 50 milliseconds for real-time uses like search, ads, and fraud checks.

@liquidai on X

Liquid AI released d1-ultra, its most powerful decision model, which handles images and text in 44 languages.

@liquidai on X

Liquid AI released LFM2.5-24B-A2B, a 24B reasoning model with 131K context that runs locally on laptops and PCs, with open weights.

@liquidai on X

Liquid AI released Open d1, two open-weight decision models: d1-3B for text and vision, and d1-omni-600M for text with image or audio.

@liquidai on X

Liquid AI added image input to its d1 decision model, so it now accepts images, text, or both through the Liquid API.

@liquidai on X

Liquid AI announced d1, its first decision model, built for fast structured decisions in software and available on the Liquid API.

@liquidai on X

Liquid AI released an experimental DSpark draft model for LFM2.5-VL-3B that speeds up vision-language decoding by up to 3.13x on Apple M5 Max.

@liquidai on X

Liquid AI optimized Liquid Context, its on-device context layer for AI agents, for Snapdragon processors, and it now runs on the Qualcomm Hexagon NPU.

@liquidai on X

Liquid AI released LFM2-1.2B-Longevity and LFM2-2.6B-Longevity, two compact models for aging research, plus the LongevityBench test suite, now available on Hugging Face.

@liquidai on X

Liquid AI released Pipette, an open source suite with Artificial Analysis for measuring AI model quality, speed, latency and memory on phones, laptops and other devices.

@liquidai on X