Wispr Flow - Don’t type, just speak
Mini toolsLFM2.5 VL 450M
By Liquid AI
LFM2.5-VL-450M is an upgraded 450M-parameter vision-language model from Liquid AI designed for low-latency, production-grade visual reasoning. Liquid says it improves on LFM2-VL-450M by scaling pretraining from 10T to 28T tokens and adding post-training focused on grounding, instruction following, and reliability in multimodal deployments. It adds bounding-box prediction, supports prompts in eight languages, and includes function calling, while remaining fast enough for edge devices such as Jetson Orin, phones, and compact PCs. Liquid positions it for industrial automation, always-on monitoring, retail, and e-commerce workflows.
New
Image
Gen 4
Overview
LFM2.5-VL-450M is Liquid AI’s compact vision-language model for structured visual intelligence from edge to cloud. It is built to turn image streams into grounded, actionable outputs in real time, adding object grounding, better instruction following, multilingual image understanding, and function calling support while staying efficient enough for edge hardware.
About Liquid AI
Liquid AI is an MIT spin-off building efficient general-purpose AI models (Liquid Foundation Models, or LFMs) that run on edge devices with less memory and power.
They recently raised $250M in Series A funding to scale model development and deployment.
Last updated: April 9, 2026
