Open Source Mobile Operating Systems Large Language Models (LLM)

DeepSeek-V3 is a robust Mixture-of-Experts (MoE) language model developed by DeepSeek, featuring a total of 671 billion parameters, with 37 billion activated per token. It employs Multi-head Latent Attention (MLA) and the DeepSeekMoE architecture to enhance computational efficiency. The model introduces an auxiliary-loss-free load balancing strategy and a multi-token prediction training objective to boost performance. Trained on 14.8 trillion diverse, high-quality tokens, DeepSeek-V3 underwent supervised fine-tuning and reinforcement learning to fully realize its capabilities. Evaluations indicate that it outperforms other open-source models and rivals leading closed-source models, achieving this with a training duration of 55 days on 2,048 Nvidia H800 GPUs, costing approximately $5.58 million.

DeepSeek-R1 is an open-source large language model developed by DeepSeek, designed to excel in complex reasoning tasks across domains such as mathematics, coding, and language. DeepSeek R1 offers unrestricted access for both commercial and academic use. The model employs a Mixture of Experts (MoE) architecture, comprising 671 billion total parameters with 37 billion active parameters per token, and supports a context length of up to 128,000 tokens. DeepSeek-R1's training regimen uniquely integrates large-scale reinforcement learning (RL) without relying on supervised fine-tuning, enabling the model to develop advanced reasoning capabilities. This approach has resulted in performance comparable to leading models like OpenAI's o1, while maintaining cost-efficiency. To further support the research community, DeepSeek has released distilled versions of the model based on architectures such as LLaMA and Qwen.

nndeploy is an open-source framework designed to simplify the deployment of artificial intelligence models across multiple hardware platforms and devices. The framework focuses on making it easier to transform trained AI models into production-ready applications that can run efficiently on desktops, mobile devices, servers, and edge computing hardware. Developers can use visual workflows to design and configure AI processing pipelines by connecting modular nodes that represent different stages of the inference process. The system supports multiple inference engines and hardware accelerators, allowing the same AI workflow to run on different platforms without significant modifications. nndeploy also includes performance optimization techniques such as parallel execution, memory reuse, and hardware-accelerated operations to improve inference speed.

Fullmoon is a free, open-source application that enables users to interact with large language models directly on their devices, ensuring privacy and offline accessibility. Optimized for Apple silicon, it operates seamlessly across iOS, iPadOS, macOS, and visionOS platforms. Users can personalize the app by adjusting themes, fonts, and system prompts, and it integrates with Apple's Shortcuts for enhanced functionality. Fullmoon supports models like Llama-3.2-1B-Instruct-4bit and Llama-3.2-3B-Instruct-4bit, facilitating efficient on-device AI interactions without the need for an internet connection.

RWKV (pronounced as RwaKuv) is an RNN with GPT-level LLM performance, which can also be directly trained like a GPT transformer (parallelizable). So it's combining the best of RNN and transformer - great performance, fast inference, fast training, saves VRAM, "infinite" ctxlen, and free text embedding. Moreover it's 100% attention-free. Default configs has enabled custom CUDA kernel acceleration, which is much faster and consumes much less VRAM. If you encounter possible compatibility issues, go to the Configs page and turn off Use Custom CUDA kernel to Accelerate.

Run a fast ChatGPT-like model locally on your device. This combines the LLaMA foundation model with an open reproduction of Stanford Alpaca a fine-tuning of the base model to obey instructions (akin to the RLHF used to train ChatGPT) and a set of modifications to llama.cpp to add a chat interface. Download the zip file corresponding to your operating system from the latest release. The weights are based on the published fine-tunes from alpaca-lora, converted back into a PyTorch checkpoint with a modified script and then quantized with llama.cpp the regular way.

AIdea is a comprehensive Flutter-based cross-platform app integrating major AI models—OpenAI GPT, Chinese models Tongyi Qianwen and Wenxin Yiyan, plus image models like Stable Diffusion for text-to-image, image-to-image, SDXL 1.0, super-resolution, and colorization. It includes a client app, server backend, and Docker deployment scripts for hosted setups.

Cake is a compact, powerful toolkit that combines a flexible TCP/UDP proxy, port forwarding system, and connection manager designed for both development and penetration testing scenarios. It enables users to create complex networking flows where traffic can be proxied, relayed, and manipulated between endpoints — useful for debugging networked applications, inspecting protocols, or tunneling traffic through different hops. The tool is designed to work with multiple protocols and supports dynamic rule definitions so that incoming and outgoing connections can be routed, rewritten, or logged according to user-defined policies. Unlike many simple proxies, Cake can act as a full connection broker: it can bind to arbitrary interfaces, handle simultaneous upstream/downstream sessions, and apply traffic rules on the fly. This makes it suitable for troubleshooting tricky network behavior, simulating network conditions, or chaining services in a modular test environment.

Doctor Dignity is a prototype project exploring how AI-assisted tooling might support compassionate, accessible health guidance for people who struggle to get timely care. The repository centers on a simple end-to-end pipeline—intake of user-reported symptoms, basic triage logic, and clear, supportive messaging—intended to demonstrate how such systems could be built. It emphasizes a humane UX: plain-language prompts, de-jargonized outputs, and guardrails that nudge users toward professional care when needed. The code is designed to be hackable rather than production-grade, giving learners a chance to experiment with NLP flows and lightweight back-end components. It also highlights privacy-aware patterns and cautions that this kind of software must not replace licensed medical advice. As a teaching and ideation vehicle, the project invites contributors to iterate on intent classification, response templates, and safe-use boundaries.