# Open-Source Model

HiDream-I1 is a new open-source image generation base model with 17 billion parameters, capable of generating high-quality images within seconds. This model is suitable for research and development and has performed excellently in multiple evaluations, demonstrating high efficiency and flexibility, making it suitable for various creative design and generation tasks.

Together Chat is a secure AI chat platform offering 100 free messages per day, ideal for users who need private conversations and high-quality interactions. It uses North America servers to ensure user information security.

Wan 2.1 AI is an open-source large-scale video generation AI model developed by Alibaba. It supports text-to-video (T2V) and image-to-video (I2V) generation, capable of transforming simple input into high-quality video content. This model is significant in the field of video generation, greatly simplifying the video creation process, lowering the creation threshold, improving creation efficiency, and providing users with a wide variety of video creation possibilities. Its main advantages include high-quality video generation effects, smooth presentation of complex actions, realistic physical simulation, and rich artistic styles. Currently, this product is fully open-source, and users can use its basic functions for free. It is highly practical for individuals and businesses that have video creation needs but lack professional skills or equipment.

CSM 1B is a speech generation model based on the Llama architecture, capable of generating RVQ audio codes from text and audio input. The model is primarily used in speech synthesis and boasts high-quality speech generation capabilities. Its advantages include the ability to handle multi-speaker dialogue scenarios and generate natural and fluent speech through contextual information. This open-source model is intended to support research and educational purposes but is explicitly prohibited from being used for impersonation, fraud, or illegal activities.

Wan2.1-T2V-14B is an advanced text-to-video generation model based on a diffusion transformer architecture, incorporating innovative spatiotemporal variational autoencoders (VAEs) and large-scale data training. It generates high-quality video content at various resolutions, supports both Chinese and English text input, and surpasses existing open-source and commercial models in performance and efficiency. This model is suitable for scenarios requiring efficient video generation, such as content creation, advertising production, and video editing. Currently, this model is freely available on the Hugging Face platform to promote the development and application of video generation technology.

Wan is an advanced visual generation model developed by Alibaba's DAMO Academy, boasting powerful video generation capabilities. It can generate videos based on text, images, and other control signals. The Wan2.1 series models are now fully open-sourced. Key advantages include: exceptional complex motion generation, producing realistic videos with a wide range of body movements, complex rotations, dynamic scene transitions, and smooth camera movements; accurate physics simulation, generating videos that adhere to real-world physics; cinematic-quality visuals, offering rich textures and diverse stylistic effects; and controllable editing capabilities, supporting precise editing using image or video references. The open-sourcing of this model introduces new possibilities to the video generation field, lowering the barrier to entry and driving technological advancements.

PIKE-RAG, developed by Microsoft, is a domain knowledge and reasoning-enhanced generation model designed to augment the capabilities of Large Language Models (LLMs) through knowledge extraction, storage, and inferential logic. Featuring a multi-module design, this model effectively handles complex multi-hop question answering tasks and significantly improves accuracy in industries like industrial manufacturing, mining, and pharmaceuticals. Key advantages of PIKE-RAG include efficient knowledge extraction, robust multi-source information integration, and multi-step reasoning, making it exceptionally well-suited for scenarios demanding deep domain knowledge and complex logical reasoning.

SkyReels V1 is a human-centric video generation model fine-tuned from HunyuanVideo. Trained on high-quality film clips, it generates video content with cinematic texture. The model has achieved industry-leading performance in the open-source domain, particularly excelling in facial expression capture and scene understanding. Key advantages include its open-source leadership, advanced facial animation technology, and cinematic lighting aesthetics. The model is well-suited for scenarios requiring high-quality video generation, such as film production and advertising creation, offering broad application prospects.

Llasa-1B is a text-to-speech model developed by the Audio Lab at the Hong Kong University of Science and Technology. Based on the LLaMA architecture and integrated with speech tokens from the XCodec2 codec, it converts text into natural and fluent speech. The model has been trained on 250,000 hours of Chinese and English speech data and supports generating speech from plain text, as well as utilizing given voice prompts for synthesis. Its main advantage is the ability to produce high-quality multilingual speech, making it suitable for various applications such as audiobooks and voice assistants. The model is licensed under CC BY-NC-ND 4.0, prohibiting commercial use.

OLMo-2-1124-7B-Instruct is a large language model developed by the Allen Institute for AI, focusing on dialogue generation tasks. This model has been optimized for various tasks including mathematical problem-solving, GSM8K, IFEval, and has undergone supervised fine-tuning on the Tülu 3 dataset. It is built on the Transformers library and can be used for research and educational purposes. The main advantages of the model include high performance, multi-task adaptability, and being open-source, making it an essential tool in the realm of natural language processing.

Aria is a multimodal native mixture of experts model that excels in multimodal, language, and coding tasks. It performs exceptionally well in video and document understanding, supporting up to 64K multimodal input, with the ability to describe a 256-frame video in just 10 seconds. The model has 25.3 billion parameters and can be loaded on a single A100 (80GB) GPU using bfloat16 precision. Aria was developed to meet the needs for multimodal data understanding, particularly in video and document processing. It is an open-source model aimed at advancing multimodal artificial intelligence.

DeepSeek-Coder-V2 is an open-source Mixture-of-Experts code language model, whose performance rivals that of GPT4-Turbo. It excels in code-specific tasks. Further pre-trained with an additional 60 billion tokens, it enhances coding and mathematical reasoning abilities while maintaining comparable performance on general language tasks. Compared to DeepSeek-Coder-33B, it shows significant improvements across code-related tasks, reasoning, and general capabilities. Moreover, it supports 338 programming languages (an expansion from 86) and boasts a context length extended to 128K from 16K.

DeepSeek-Coder-V2 is an open-source Mixture-of-Experts (MoE) code language model with performance comparable to GPT4-Turbo, showcasing exceptional performance in code-specific tasks. Built upon DeepSeek-Coder-V2-Base, it has undergone further pre-training using a high-quality, multi-source corpus of 6 trillion tokens. This has significantly enhanced its coding and mathematical reasoning capabilities while maintaining its performance on general language tasks. Supported programming languages have expanded from 86 to 338, and the context length has increased from 16K to 128K.

SiliconCloud is a high-performance GenAI cloud service based on excellent open-source foundation models. Its main advantages include a fast model access experience, diverse model services, and a simple and easy-to-use development interface. SiliconCloud aims to provide users with high-quality and low-cost AI model services.

Stable Audio Open is an open-source text-to-audio model optimized for generating short audio samples, sound effects, and production elements. It allows users to generate up to 47 seconds of high-quality audio data using simple text prompts, particularly suitable for creating percussion hits, instrument improvisations, environmental sounds, foley recordings, and more for music production and sound design. A key benefit of open-sourcing is that users can fine-tune the model with their own customized audio data.

360Zhinao is a series of 7B-scale intelligent language models open-sourced by Qihoo 360, including base models and three dialogue models with different context lengths. These models have been pre-trained on massive Chinese and English language datasets and have demonstrated excellent performance in various tasks such as natural language understanding, knowledge, mathematics, code generation, and have strong capabilities in long-text conversations. The model can be used in the development and deployment of various conversational applications.