# Lightweight deployment

Quantized version of Qwen2.5-VL-7B-Meteorology, suitable for image-text processing tasks related to meteorology.

Mi:dm 2.0 is an AI model centered around South Korea, developed by KT using its proprietary technology. This model has deeply internalized the unique values, cognitive frameworks, and common-sense reasoning of South Korean society. It can not only process and generate Korean content but also reflect a profound understanding of South Korean social and cultural norms and values.

The quantized version of Tencent Hunyuan A13B Instruction Model, which uses technical means to improve operational efficiency while ensuring performance.

Apollo2-7B-GGUF is a quantized version of FreedomIntelligence/Apollo2-7B, supporting medical large language model applications in multiple languages.

This is a 4-bit DWQ quantized version converted from Qwen/Qwen3-Embedding-8B, suitable for the embedding model of the MLX framework.

PP-OCRv4_mobile_det is an efficient text detection model optimized for mobile devices developed by the PaddleOCR team, suitable for deployment on edge devices.

The quantized version of Qwen3-Reranker-0.6B, dedicated to making knowledge accessible to everyone.

PP-OCRv5_mobile_det is the latest generation of lightweight text detection model developed by the PaddleOCR team, supporting efficient text detection in multiple languages and scenarios.

This project provides optimized GGUF quantized files, which can significantly improve model performance. These quantized files are supported by Featherless AI. Users can run any desired model by paying a small fee.

This project is a quantized version of kakaocorp/kanana-safeguard-8b, dedicated to making knowledge accessible to the public.

This is an 8-bit quantized version in MLX format converted from the DeepSeek-R1-0528-Qwen3-8B model, suitable for text generation tasks.

This model is a 4-bit quantized version converted from DeepSeek-R1-0528-Qwen3-8B, optimized for the MLX framework and suitable for text generation tasks.

Qwen2.5-Omni-7B-GGUF is the GGUF format version of the Qwen2.5-Omni-7B model, supporting multimodal inputs including text, audio, and images.

This project provides a quantized version of academic-ds-9B, aiming to make knowledge accessible to everyone.

Devstral-Small-2505-8bit is an 8-bit quantized model converted from mistralai/Devstral-Small-2505, suitable for the MLX framework and supporting text generation tasks in multiple languages.

Skywork-OR1-7B is a 7B-parameter large language model offering multiple quantization versions to accommodate different hardware requirements.

This model is a 4-bit DWQ quantized version of Qwen3-4B, converted to the MLX format for easy text generation using the mlx library.

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

The quantized version of Huihui AI's Qwen3-4B model, aiming to make knowledge more widely accessible to the public.

Phi-4-mini-reasoning is a lightweight open model built on synthetic data, focusing on high-quality, reasoning-rich data, and further fine-tuned for more advanced mathematical reasoning capabilities.

An optimized 8-bit quantized version of Qwen3-8B, designed for efficient inference on the MLX framework

This is a 6-bit quantized version of the Qwen3-4B model converted to the MLX format, suitable for text generation tasks.

Qwen3-8B-4bit-DWQ is a 4-bit quantized version of Qwen/Qwen3-8B converted to the MLX format, optimized for efficient operation on Apple devices.

This is the ONNX version of the AST (Audio Spectrogram Transformer) model, designed specifically for audio classification tasks and fine-tuned on the AudioSet dataset.

This is a quantized version of the Microsoft Phi - 4 - mini - reasoning model, which is quantized using the llamacpp tool to improve the model's operating efficiency and performance in different hardware environments.

This model is a GGUF format text-to-speech model converted from MYZY-AI/Muyan-TTS-SFT, supporting Chinese speech synthesis.

Foundation-Sec-8B is a large language model based on the Transformer architecture, focusing on text generation tasks.

An instruction fine-tuned model optimized based on the Mistral architecture, suitable for zero-shot classification tasks

This is the 4-bit quantized version of the Qwen/Qwen3-8B model, converted to the MLX framework format, suitable for efficient inference on Apple silicon devices.

Qwen3-4B-4bit is a 4-bit quantized version converted from Qwen/Qwen3-4B to the MLX format, designed for efficient operation on Apple chips.

The 4-bit quantized version of the MNN model for Qwen3-4B, used for efficient text generation tasks

A 4-bit quantized version based on InternVL2_5-1B, suitable for text generation and chat scenarios.

A 3B-parameter language model based on the Llama architecture, offering multiple quantization versions to suit different hardware needs

Mistral-Small-24B-Instruct-2501 is a 24B-parameter instruction-finetuned large language model supporting multilingual text generation tasks.

Gemma 3 is a lightweight, state-of-the-art multimodal open-source model launched by Google, capable of processing text and image inputs and generating text outputs.

Gemma 3 is a lightweight multimodal model launched by Google, capable of processing text and image inputs and generating text outputs. This model has a 128K large context window and multilingual support for over 140 languages.

A Korean-optimized text generation model with instruction-following capability, featuring lightweight design suitable for edge device deployment

Gemma 3 is a lightweight, advanced open model series from Google, built on the same research and technology used to create Gemini models. This model is multimodal, capable of processing both text and image inputs to generate text outputs.

GigaAM v2 is an automatic speech recognition (ASR) model that supports Russian speech-to-text tasks, offering both CTC and RNN-T architectures.

Gemma is a lightweight open model series launched by Google, built on Gemini model technology. Gemma 3 is a multimodal model supporting text and image inputs with text outputs, featuring a 128K large context window and multilingual capabilities.