Setup your First Server

Open-Sora

Open-Sora-v2 is an open-source Text-to-Video generation model developed by HPC-AI Technology. The project aims to democratize high-quality video creation by offering a publicly accessible alternative to closed systems like OpenAI's Sora. It is built upon the Diffusion Transformer (DiT) architecture and optimized using frameworks like ColossalAI for high computational efficiency, crucial for training and inference. Key features include generating dynamic video clips from text prompts (Text-to-Video) or a source image (Image-to-Video). It utilizes components like StabilityAI's VAE for video compression and T5/CLIP for robust text and image encoding. The model prioritizes scalability and accessibility, making advanced video synthesis available to researchers and users with significant, but non-exclusive, computing resources.

Kimi K2

Kimi K2 is our latest Mixture-of-Experts model with 32 billion activated parameters and 1 trillion total parameters. It achieves state-of-the-art performance in frontier knowledge, math, and coding among non-thinking models. But it goes further — meticulously optimized for agentic tasks, Kimi K2 does not just answer; it acts. And now, it is within your reach. Today, we are open-sourcing: Kimi-K2-Base: The foundation model, a strong start for researchers and builders who want full control for fine-tuning and custom solutions. Kimi-K2-Instruct: The post-trained model best for drop-in, general-purpose chat and agentic experiences. It is a reflex-grade model without long thinking. With Kimi K2, advanced agentic intelligence is more open and accessible than ever. We can't wait to see what you build.

Phi

Phi is a family of small, open-weight language models developed by Microsoft. They are designed to be compact, fast, and efficient for local or edge use. Models like Phi-2 (2.7B) and Phi-3-mini (3.8B) offer strong performance on reasoning and coding tasks despite their small size. Phi uses a mix of curated web data, academic texts, and synthetic datasets. These models run well on laptops and devices with limited memory, especially when quantized. Developers use Phi for fast prototypes, embedded AI tools, or private inference.

Gemma

Google Gemma is a family of open-weight language models designed for efficient reasoning and high-quality text generation. Built on the same research foundation as Gemini, it offers strong performance while remaining lightweight and privacy-friendly. Gemma models come in multiple sizes, such as 1B, 2B, and 9B, optimized for chat, coding, and multilingual tasks. They run efficiently on local devices like laptops or Mac mini, making them ideal for open and private AI applications.

FLUX

FLUX is a new family of image generation models by Black Forest Labs, built on Flow Matching Transformer technology. It produces highly realistic, detailed, and coherent images from text prompts with faster inference than diffusion models. Versions like FLUX.1-dev focus on quality, while FLUX.1-schnell prioritizes speed and efficiency. It’s optimized for both local setups and professional workflows, supporting tools like ComfyUI and Hugging Face Diffusers.