Qwen3-VL is the latest generation of vision-language model launched by Alibaba, with significant improvements in text understanding, visual perception, spatial understanding, video analysis, and agent interaction. This model supports multimodal input and has powerful reasoning ability and long context processing ability.

MultimodalTransformers

PokeeResearch - 7B is a 7 - billion - parameter in - depth research agent model developed by Pokee AI. It combines Reinforcement Learning from AI Feedback (RLAIF) and a powerful reasoning framework, enabling reliable, aligned, and scalable research - level reasoning in tool - enhanced large language models. It is suitable for complex multi - step research workflows.

Natural Language ProcessingTransformersEnglish

This is a quantized version of the Alibaba Tongyi Deep Research 30B-A3B model. It uses MXFP4_MOE quantization technology and additionally adds imatrix quantization, aiming to optimize model performance and resource utilization efficiency, and is suitable for text generation tasks.

Natural Language ProcessingGguf

Qwen-Image-Edit-MeiTu is an improved version based on Qwen-Image-Edit. It is fine-tuned using the DiT architecture and focuses on enhancing visual consistency, aesthetic quality, and structural alignment capabilities in complex image editing.

Computer VisionDiffusersMultiple Languages

Based on Stable Diffusion XL, the UNet weights are fine-tuned using the Inversion-DPO method, combining direct preference optimization (DPO) technology and the inversion method to improve the quality and alignment of image generation in the diffusion model.

Computer VisionDiffusers

Qwen3-VL-8B-Thinking is an 8-billion-parameter multimodal large language model developed by the Qwen team at Alibaba Cloud. It is designed for deep multimodal reasoning, supporting visual understanding, long context processing, and structured chain-of-thought generation, and performs excellently in complex reasoning tasks.

MultimodalMlx

Qwen3-VL is the latest generation of vision-language model launched by Alibaba. It has excellent text understanding and generation capabilities, in-depth visual perception and reasoning abilities, supports long context processing and video understanding, and provides an instruction-optimized version.

MultimodalSafetensors

This project distills the inference ability of GPT into the Llama-3.1-8B model through an innovative two-stage training process. First, knowledge distillation and format alignment are performed through supervised fine-tuning, and then reinforcement learning is used to encourage the model to independently explore and optimize inference strategies, focusing on breakthroughs in the field of mathematical reasoning.

Natural Language ProcessingTransformersMultiple Languages

G²RPO is a novel reinforcement learning framework specifically designed for preference alignment in flow models, significantly improving the generation quality through a granular reward evaluation mechanism.

MultimodalDiffusers

Granite-4.0-H-Small is a long-context instruction model developed by IBM with 32 billion parameters, fine-tuned based on Granite-4.0-H-Small-Base. This model combines open-source instruction datasets and internal synthetic datasets, and uses techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has significantly improved instruction following and tool invocation capabilities, and is particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-H-Micro is a 3-billion parameter long-context instruction model developed by IBM, fine-tuned from Granite-4.0-H-Micro-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has a structured chat format and performs excellently in instruction following and tool invocation capabilities.

Natural Language ProcessingTransformers

Granite-4.0-H-Tiny is a long-context instruction model developed by IBM with 7 billion parameters, fine-tuned based on Granite-4.0-H-Tiny-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has enhanced instruction-following and tool invocation capabilities, making it particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-Micro is a long-context instruction model with 3 billion parameters developed by IBM, fine-tuned based on Granite-4.0-Micro-Base. This model combines open-source instruction datasets and internal synthetic datasets, and is developed using technologies such as supervised fine-tuning, reinforcement learning alignment, and model merging. It has enhanced instruction-following and tool invocation capabilities, and is particularly suitable for enterprise-level applications.

Natural Language ProcessingTransformers

Granite-4.0-H-Small is a long-context instruction model with 32 billion parameters developed by IBM, fine-tuned based on Granite-4.0-H-Small-Base. This model combines open-source instruction datasets and internal synthetic datasets and is developed using techniques such as supervised fine-tuning, reinforcement learning alignment, and model merging, with significant improvements in instruction following and tool invocation capabilities.

Natural Language ProcessingTransformers

This is a text-to-video LoRA model based on AI technology, specifically optimized and trained for the Alita character. It can be used in scenarios such as image generation, providing convenience for applications in related fields.

MultimodalDiffusers

This model is based on GPT-OSS-20B and fine-tuned using the Unsloth reinforcement learning framework. The aim is to optimize inference efficiency and reduce vulnerabilities that occur during reinforcement learning from human feedback (RLHF) training. The fine-tuning process focuses on the robustness and efficiency of alignment, ensuring that the model maintains inference depth without incurring excessive computational overhead.

Natural Language ProcessingTransformersEnglish

This is the GGUF quantized version of Alibaba's Qwen series large language model with 8 billion parameters, designed for advanced reasoning, intelligent agent behavior, and multilingual tasks, and can be used with llama.cpp and compatible tools.

Natural Language ProcessingGgufMultiple Languages

This project is the A3B quantized version of Alibaba's Tongyi In-depth Research 30B model. It aims to reduce the model deployment cost through quantization technology and make knowledge accessible to everyone. This model is optimized based on a large language model with a scale of 30B parameters, maintaining the strong capabilities of the original model while improving the inference efficiency.

Natural Language ProcessingGguf

Qwen2.5-VL-7B-Instruct is a multimodal vision-language model developed by the Tongyi Qianwen team at Alibaba. Based on a parameter scale of 7 billion, it is specifically optimized and trained for visual question-answering tasks. This model can understand and analyze image content and generate accurate natural language answers.

MultimodalSafetensorsEnglish

Alisia-7B-it is a quantized version based on the Gems234/Alisia-7B-it base model, supporting multiple tasks such as text generation inference, question answering, fact-checking, and reasoning. This project provides multiple quantized versions to meet the usage requirements in different scenarios.

Natural Language ProcessingTransformersMultiple Languages