Back to News
Liquid AI•August 12, 2026
LFM2.5-VL-3B
Open Source
Vision-Language Model
Edge AI
On-device
Open-weight
### TL;DR
LFM2.5-VL-3B is a 3.1-billion-parameter vision-language model designed for high-performance, on-device inference. It features significant improvements in screen understanding, grounding, and function calling, allowing it to outperform larger models while maintaining low latency on edge hardware.
Key Insights & Metrics
Pricing
Open-Weight
Cost structure
Version
2.5
Current release version
Hardware
Compatible with Apple Silicon, AMD, Qualcomm, and NVIDIA hardware; runs on-device with approximately 3 GB of memory
Compute requirements
Category
Open Source
Licensing model
Region
United States
Primary region
Key Features
- Advanced screen and UI understanding for mobile, web, and desktop
- Enhanced function calling and tool-use capabilities
- Improved grounding performance with high-precision bounding box detection
- Multi-image input reasoning support
- Optimized for edge inference with support for llama.cpp, MLX, and vLLM
- Non-reasoning architecture for low-latency, direct responses
Discussion
0
Upvotes
0
Downvotes
0 reviews
Sign in to leave a review
Reviews
No reviews yet. Be the first to review!