Liquid AILFM2.5
LFM2.5-VL-450M
This compact vision model pairs LFM2.5-350M with an 86M SigLIP2 NaFlex encoder.
What the release provides
Its 32,768-token context supports visual instruction following and object locations. Large images are split into 512-pixel tiles with a thumbnail for global context. Text-only function calls are also supported.LFM2.5-VL-450M official model card
How to use it well
Practical guidance. Check small text and distant objects at the chosen tile budget. Increasing the image resolution can increase processing cost without adding useful information.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | LiquidAI/LFM2.5-VL-450M |