Neptun3 One
A 350M open-weight model with a hybrid architecture, native tool use, and quantizations that run on your own hardware. 350M parameters.
218 downloads
ollama run hf.co/Wolkyen/Neptun3-One:Q4_K_MA 350M open-weight model with a hybrid architecture, native tool use, and quantizations that run on your own hardware. 350M parameters.
218 downloads
ollama run hf.co/Wolkyen/Neptun3-One:Q4_K_MOpen Weights. Small Models. Big Possibilities.
Meet Neptun3, Wolkyen's open-weight AI model series, built to make powerful, efficient artificial intelligence more accessible.
Designed for developers, enthusiasts, and everyday applications, Neptun3 brings together compact architectures, efficient inference, and practical intelligence without requiring massive computing resources.
With open weights at its core, Neptun3 gives developers the freedom to download, run, modify, and build with AI on their own hardware.
350 Million Parameters. Hybrid Architecture. Native Tool Use.
Neptun3 One marks the beginning of the Neptun3 series.
With just 350 million parameters, Neptun3 One is designed to challenge models in its size class, delivering capable general-purpose assistance, lightweight mathematical reasoning, and native tool-calling support.
Its hybrid architecture balances performance and efficiency, making it suitable for local applications, lightweight agents, and resource-constrained environments.
Neptun3 One is available to download through Hugging Face, with multiple quantization options for different hardware requirements.
Get started with Ollama: ollama run hf.co/Wolkyen/Neptun3-One:Q4_K_M
Download Neptun3 One: https://huggingface.co/Wolkyen/Neptun3-One
Neptun3 One is only the beginning.
The Neptun3 series will continue exploring how much capability can fit into efficient, accessible AI models, pushing the boundaries of what smaller architectures can achieve.
Neptun3 — Intelligence Without the Weight.
Developed by Wolkyen.
No reviews yet
Shared with everyone. No account.