# Lightweight Deployment

Mi:dm 2.0 is an 'AI centered around South Korea' model developed using KT's proprietary technology, which deeply internalizes the unique values, cognitive frameworks, and common-sense reasoning of South Korean society.

This is a statically quantized version based on the Qari-OCR-0.3-SNAPSHOT-VL-2B-Instruct-merged model, mainly used for image-to-text conversion tasks.

An efficient language model specifically designed for software engineering projects, featuring a lightweight design and supporting a 128k large context window, suitable for complex coding tasks.

Cosmos-Reason1-7B is a 7B-parameter foundational model released by NVIDIA, specializing in image-to-text tasks.

Quantized version of Devstral-Small-2505, offering multiple precision options to adapt to different hardware requirements

Devstral-Small-2505 is a small language model based on the Mistral architecture, supporting text generation tasks and capable of basic visual functions through compatible mmproj files.

Devstral is a large language model agent for software engineering tasks developed by Mistral AI in collaboration with All Hands AI, excelling in exploring codebases with tools, editing multiple files, and driving software engineering agents.

Devstral is an intelligent LLM specifically designed for software engineering tasks, jointly developed by Mistral AI and All Hands AI. It excels in code exploration, multi-file editing, and driving software engineering agents.

Devstral is an intelligent large language model specifically designed for software engineering tasks, jointly developed by Mistral AI and All Hands AI. It excels in code exploration, multi-file editing, and driving software engineering agents.

Devstral is a large language model for software engineering task agents, developed in collaboration between Mistral AI and All Hands AI. It excels at using tools to explore codebases, edit multiple files, and drive software engineering agents.

Devstral is an intelligent large language model specifically designed for software engineering tasks, developed in collaboration by Mistral AI and All Hands AI. It excels in codebase exploration, multi-file editing, and driving software engineering agents.

Devstral is an intelligent large language model specifically designed for software engineering tasks, jointly developed by Mistral AI and All Hands AI. It excels in code exploration, editing, and driving software engineering agents.

Static quantized version of Sam-reason-S2.1, offering multiple quantization options to suit different hardware requirements

A multimodal model fine-tuned based on Qwen/Qwen2-VL-2B-Instruct, optimized for OCR, image-to-text conversion, LaTeX math solving, and handwriting recognition

This model is a GGUF format conversion of llava-hf/llava-1.5-7b-hf, supporting image-to-text generation tasks.

TEN VAD is a low-latency, lightweight, and high-performance streaming voice activity detection system, suitable for real-time voice processing scenarios.

Devstral is an intelligent large language model developed by Mistral AI in collaboration with All Hands AI for software engineering tasks, excelling in codebase exploration, multi-file editing, and driving software engineering agents.

INTELLECT-2-GGUF is the GGUF format quantized version of PrimeIntellect/INTELLECT-2, suitable for text generation tasks.

ACE-Step-v1-3.5B is a text-to-audio model that supports high-quality audio generation, suitable for music and sound effects creation.

This is a model based on the 🤗 transformers library, with specific functions and purposes not yet clearly stated.

OpenVision is a fully open, cost-effective family of advanced vision encoders designed for multimodal learning.

OpenVision is a fully open, cost-effective family of advanced visual encoders focused on multimodal learning.

OpenVision is a fully open, cost-effective advanced visual encoder family focused on multimodal learning.

This is an automatic speech recognition model that has been converted to a version suitable for MLX and can perform inference quickly.

OLMo-2-0425-1B-Instruct is a 1-billion-parameter instruction-finetuned language model developed by AllenAI, focused on text generation tasks.

Quantized version of Qwen3-4B-abliterated, quantized using llama.cpp, supports multiple quantization types, suitable for text generation tasks.

This is the GGUF quantized version of the Josiefied-Qwen3-4B-abliterated-v1 model, suitable for local deployment and execution.

Dia is a 1.6 billion parameter open-source text-to-speech model that supports highly realistic dialogue and non-verbal expression generation

This repository contains GGUF format model files of JungZoona/T3Q-qwen2.5-14b-v1.0-e3, quantized by TensorBlock's machine and compatible with llama.cpp.

Dia is an open-weight text-to-dialogue model that supports dialogue text generation and speech synthesis.

GLM-4-9B-0414-abliterated is a large language model with 9B parameters based on the GLM architecture, suitable for text generation tasks.

A quantized version of Google's Gemma 3B model based on QAT weights, supporting multiple quantization levels for efficient inference in resource-constrained environments.

This is a multimodal model based on the Llama architecture, supporting vision and text instructions, optimized with 4-bit quantization.

A merged model based on Meta-Llama-3.1-8B-Instruct, fine-tuned for legal summarization tasks, capable of converting legal terminology into concise and understandable summaries.

This model is a GGUF format model converted from the IBM Granite-3.3-8B instruction fine-tuned model, suitable for text generation tasks.

An 8-bit quantized version converted from the Google Gemma 3 12B model, suitable for image-text to text tasks.

This is an MLX-format model converted from the Google Gemma 3-12B model, supporting image-text-to-text tasks.

Salesforce's 800M parameter Llama-xLAM-2 model quantized version, specialized in text generation tasks

Gemma 3 is a lightweight open-source multimodal model series developed by Google, built on Gemini technology, supporting text and image inputs with text outputs. The 1B version has undergone instruction tuning and quantization-aware training (QAT), making it suitable for deployment in resource-constrained environments.

GLM-4-Z1-9B-0414 is the latest open-source model in the GLM family, featuring excellent mathematical reasoning and general capabilities, suitable for lightweight deployment in resource-constrained scenarios.