TAAFT
Free mode
100% free
Freemium
Free Trial
Prompts Deals
Companies
Liquid AI
Models
LFM2.5 VL 450M Extract

LFM2.5 VL 450M Extract

By Liquid AI
Model family: LFM
LFM2.5-VL-1.6B-Extract is the larger Liquid AI image-extraction model in the same Extract line. It extracts user-defined fields from images and returns valid JSON, extending Liquid’s Extract family from text-document extraction into visual data. Liquid reports strong benchmark numbers for JSON validity, field F1, and VLM-judge score, positioning the 1.6B version as a higher-capability option for structured image extraction workflows.
Multimodal
Released: April 8, 2026

Overview

LFM2.5-VL-1.6B-Extract is Liquid AI’s 1.6B vision-language extraction model for image-to-JSON structured field extraction.

🔍Data extraction🔍Image interpretation📜OCR

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.

Industry: Artificial Intelligence
Company Size: 105
Location: Cambridge, MA, US
Website: liquid.ai
View Company Profile

Other models from this family

View all models from this family

Related Models

  • GLM 4.5V
    GLM-4.5V is a 106B parameter (12B active) MoE-based visual reasoning model supporting image, video, document, and GUI task understanding. It achieves open-source SOTA performance across 42 visual multimodal benchmarks and features a switchable Thinking Mode for balancing response speed against deep reasoning depth.
    Multimodal
    Released 11mo ago
  • MiniCPM-o 4.5
    MiniCPM-o 4.5 is an on-device multimodal LLM (~9B params) that matches Gemini 2.5 Flash on vision and speech, supporting full-duplex live streaming so it can see, listen and speak in real time.
    🔊Text to speech🗂️Flashcards🎥Videos
    Multimodal
    Released 10mo ago
  • ACE-Step 1.5 XL (4B DiT)
    ACE-Step 1.5 XL is the 4B-parameter Diffusion Transformer decoder in the ACE-Step 1.5 music generation line, built for higher audio quality than the earlier 2B models. The release says it achieves the best scores across all 11 benchmark metrics, surpassing both commercial and open-source models, while remaining compatible with ACE-Step language models from 0.6B to 4B.
    🎵Music production🔊Audio🎵Songwriting
    Multimodal
    Released 3mo ago
Last updated: July 8, 2026
0 AIs selected
Clear selection
#
Name
Task