On-Device AI Optimization Platform, Opt-Hancer
OptAI provide end-to-end AI optimization pipelines to compress, accelerate, and deploy models efficiently on any device.
Model Zoo
“We provide a comprehensive range of solutions designed to help our clients effectively integrate, optimize, and leverage the full potential of these advanced language models.”
LLaMA3.2-1B & 3B
Llama 3 is a large language model developed by Meta, capable of performing a wide range of tasks.
Phi3 & 4
Microsoft’s Phi series includes the small, device-friendly Phi-3 and the more advanced, multimodal Phi-4.
HyperCLOVA X
HyperCLOVA X is a large language model from Naver, specialized in the Korean language and culture.
Qwen2-VL
Qwen2-VL is a multimodal vision-language model from Alibaba Cloud’s Qwen series, capable of understanding and analyzing both text and images.
Gemma2
Gemma 2 is the next generation of Google’s open-source language models, designed for enhanced efficiency and performance.
EXAONE-2.4B & 1.2B
Exaone is a series of large language models developed by LG AI Research, focusing on various applications across their enterprise sectors.
Midm2.0
Mi:dm 2.0 is a “Korea-centric” large language model developed by KT, known for its deep understanding of Korean culture and social context while also supporting English tasks.
Llava
LLaVA is an open-source multimodal language model that combines a vision encoder and an LLM to understand and interact with both text and images.
Qwen2.5-2B
Qwen2 is an open-source language model series developed by Alibaba Cloud, known for its strong performance and multilingual capabilities across various sizes.
Kanana-nano-2.1B
Kanana Nano is a small, open-source language model from Kakao, designed for both Korean and English.
GLM-V
GLM-V is a multimodal large language model from Zhipu AI, capable of understanding and interacting with both text and images.
Use Case #1
Voice Call Analysis
Conversational understanding via compressed LLM for accurate information extraction.
Use Case #2
Driver Monitoring
Driving monitoring via pose estimation and detection for driver state and behavior analysis.
Use Case #3
Object Detection & Tracking
Real-time detection and tracking of objects directly on the device without reliance on the cloud.
Directions
Headquarter
Opt-AI Inc.
#502, Building A 45
Yangcheong 4-gil
Cheongju, South Korea (28116)
Engineering Center
Opt-AI Inc.
#03, Superstart Lab
E10 LG Sciencepark
70 Magokjungang 10-ro
Seoul, South Korea (07795)
