Liquid AI Releases LFM2.5-VL-3B: A 3B Vision-Language Model That Reads Screens, Grounds Objects, and Calls Tools On-Device
Liquid AI launched LFM2.5-VL-3B, a 3.1B-parameter vision-language model optimized for on-device use. It improves grounding and function calling benchmarks, enabling screen reading and tool usage locally.
Liquid AI has introduced LFM2.5-VL-3B, a vision-language model designed specifically for local deployment. The architecture contains 3.1 billion parameters, targeting efficiency without sacrificing multimodal understanding capabilities.
The update introduces function calling to the vision-language line, with reported gains in tool usage benchmarks. Grounding accuracy reportedly increased substantially, while screen reading performance achieved an average score of 80.7 on standard evaluation sets.
This development highlights the trend toward compact, capable models suitable for edge computing. Local execution reduces latency and privacy concerns associated with sending visual data to remote servers for processing.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.