LFM 2.5.

A light-footed on-device pair: one for text agents that think and call tools, another for turning a camera roll, a label, or a sign into an offline conversation.

View on Hugging Face See LocalEngine variants ↓

Model makerLiquid AI

Supported in LocalEngineiPhone & iPad

Upstream licenseLFM 1.0

SourceLFM 2.5 VL model card ↗

Where it earns its space

Built for a real local rhythm.

These are good-fit starting points—not promises of benchmark performance. Choose the smallest listed option that comfortably covers your work.

Reading signs, labels, and short documents

Multilingual private chat

Tool-calling text workflows

In LocalEngine

Configured variants, by device.

The formats below are the downloads currently represented in LocalEngine’s model catalogs. GGUF runs through the llama.cpp-based runtime; MLX options use the Apple Silicon MLX runtime.

iPhone & iPad

LFM2.5 2.6B · LFM2.5-VL 1.6B

Formats: GGUF Q4_K_M / Q8_0 + Q8_0 SigLIP2 projector · MLX 4-bit / 8-bit
Local note: The 2.6B text model carries thinking and tool-call formats; VL adds image understanding, OCR, and multilingual chat.

About the source

Model identity, task category, supported formats, and license are based on the linked Hugging Face LFM 2.5 VL model card and the LocalEngine macOS/iOS catalogs. Hardware guidance reflects the packaged variants, not a universal system requirement.