INTELLECT-3 is a Mixture of Experts (MoE) inference model developed by Prime Intellect with a parameter scale of 106B (12B active parameters). It is supervised and fine-tuned based on GLM-4.5-Air-Base and trained through large-scale reinforcement learning, showing excellent performance in mathematics, coding, and inference tasks.

Natural Language ProcessingGguf

This is a CrossEncoder model developed by Redis and fine-tuned for the LangCache semantic caching task. It is based on the mature `cross-encoder/ms-marco-MiniLM-L6-v2` model and trained on a dataset of over 1 million LangCache sentence pairs. It is specifically designed to calculate the semantic relevance score between two texts to optimize the cache hit rate.

Natural Language ProcessingSentence-transformersEnglish

GigaChat3-10B-A1.8B is an efficient dialogue model in the GigaChat series. Based on the Mixture-of-Experts (MoE) architecture, it has a total of 10 billion parameters and 1.8 billion active parameters. It adopts the innovative Multi-head Latent Attention (MLA) and Multi-token Prediction (MTP) technologies, aiming to optimize inference throughput and generation speed. The model is trained on 20T tokens of diverse data and supports 10 languages including Chinese, suitable for dialogue scenarios requiring quick responses.

Natural Language ProcessingGgufMultiple Languages

Hito 1.7B is a GGUF quantized version model optimized for tools such as llama.cpp, Ollama, and LM Studio. It has 1.7 billion parameters and uses the Nested Cognitive Reasoning (NCR) method, enabling a structured and self-correcting thinking mode to improve the accuracy and transparency of reasoning.

Natural Language ProcessingGgufEnglish

This is a low-rank adapter (LoRA) for the Z-Image Turbo DiT model, specifically designed to generate images with a historical photo style. The model is trained on 150 color photos taken by photographer Sergey Prokudin-Gorsky from the 1900s to 1910s, capable of injecting the unique aesthetics of early color photography, such as warm-toned halos and specific color textures, into modern image generation.

Computer VisionDiffusers

Z-Image is an efficient basic image generation model with 6 billion parameters, specifically designed to address the efficiency and quality issues in the field of image generation. Its distilled version, Z-Image-Turbo, can reach or exceed the leading competitors with only 8 function evaluations. It can achieve sub-second inference latency on enterprise-level H800 GPUs and run on consumer-grade devices with 16G VRAM.

Computer VisionDiffusersMultiple Languages

INTELLECT-3 is a Mixture of Experts (MoE) model with 106 billion parameters, trained through large-scale reinforcement learning. It demonstrates excellent performance in mathematics, coding, and reasoning benchmark tests. The model, training framework, and environment are all open-sourced under a permissive license.

Natural Language ProcessingTransformersEnglish

This is a hybrid - language AI model specifically designed for call record analysis, capable of handling call transcription content in Hindi, English, and mixed Hindi - English. The model is fine - tuned based on Qwen2.5 - 1.5B - Instruct and has powerful multilingual understanding and information extraction capabilities.

Natural Language ProcessingTransformersMultiple Languages

This is a video understanding model fine-tuned on an unknown dataset based on the VideoMAE-base architecture, specifically designed for sign language recognition tasks. The model achieved an accuracy of 18.64% after 20 training epochs.

Computer VisionTransformers

Zenyx_114M-Tiny-Edu-Instruct is an experimental small instruction fine-tuned language model with approximately 114 million parameters. It is built on the TinyEdu-50M base model, pre-trained on the FineWeb-Edu dataset, and fine-tuned on a mixed dataset of OpenHermes-2.5 and CodeFeedback-Filtered. This model aims to explore the limits of instruction fine-tuning under a minimal architecture and verify that the loss converges to approximately 1.04.

Natural Language ProcessingPytorchEnglish

Gemma 3 is a lightweight open-source multimodal model launched by Google. This version is an instruction-tuned quantization-aware training model with 12B parameters, which has been converted to the MXFP4 format of the MLX framework. It supports text and image input and generates text output, with a 128K context window and support for over 140 languages.

MultimodalMlx

This is a video understanding model fine-tuned on an unknown dataset based on the MCG-NJU/videomae-base model. After 20 epochs of training, it achieved an accuracy of 13.31% on the evaluation set. This model is specifically optimized for video analysis tasks.

Computer VisionTransformers

The quantized GGUF version of Tencent's Hunyuan Video 1.5 model, specifically designed for text-to-video generation tasks, supports 720P HD video generation, and offers multiple quantization precision versions to optimize usage efficiency.

MultimodalGguf

This project provides a quantized GGUF version of Tencent's Hunyuan Video 1.5 text-to-video model, which supports 480P video generation tasks. It includes both distilled and full models and can be used in conjunction with tools such as ComfyUI-GGUF.

MultimodalGguf

The quantized GGUF version of Tencent's Hunyuan Video 1.5 model, specifically designed for image-to-video and video generation tasks. It offers distilled and standard models with a 480P resolution and supports multiple quantization precisions, including Q4_K_S, Q8_0, and FP16, etc.

MultimodalGguf

Strawberry_Smoothie-TEST is a hybrid model obtained by merging three 12B parameter models using the mergekit tool. It combines the advantages of Unity-12B, Chaos-Unknown-12b, and Smoothie-12B-Model_Stock, aiming to provide better text generation and dialogue capabilities.

Natural Language ProcessingTransformers

This is the quantized version of the ai-sage/GigaChat3-10B-A1.8B model, using the GGUF format and requiring support from a specific llama.cpp branch. The model adopts a Mixture of Experts architecture, with a total of approximately 11.8 billion parameters, of which approximately 1.8 billion are activated parameters.

Natural Language ProcessingGgufMultiple Languages

VibeThinker-1.5B is a 1.5-billion-parameter dense language model launched by Weibo AI. It is fine-tuned based on Qwen2.5-Math-1.5B and is specifically designed for mathematical and algorithmic coding problems. Trained using the 'Spectrum to Signal Principle' framework, it outperforms larger models in multiple math competition tests. The training cost is approximately $7,800, and it supports an output of up to about 40k tokens.

Natural Language ProcessingTransformersMultiple Languages

GigaChat3-10B-A1.8B is a dialogue model in the GigaChat series, based on the Mixture of Experts (MoE) architecture, with a total of 10 billion parameters, of which 1.8 billion are active parameters. This model uses multi-head latent attention and multi-token prediction technologies, supports a long context of 256,000 tokens, and performs excellently in multilingual dialogue and reasoning tasks.

Natural Language ProcessingSafetensorsMultiple Languages

ChronoEdit-14B is an image editing and world simulation model with temporal reasoning capabilities developed by NVIDIA, with 14 billion parameters. It achieves physically-aware image editing and action-conditioned world simulation through a two-stage reasoning process, extracting prior knowledge from pre-trained video generation models.

MultimodalDiffusersMultiple Languages