This Hugging Face repository from Unsloth hosts GGUF-quantized versions of the large Kimi K3 model, reducing its size from 1.56TB to 594GB while maintaining usability for local inference. The weights are intended for use with llama.cpp, Ollama and other GGUF-compatible runtimes, enabling developers to run this multimodal model on consumer or server hardware without relying on cloud APIs.
Kimi K3 sits in PulseGate's Foundation models & chat category. It focuses on running the very large Kimi K3 model locally by providing heavily compressed and quantized weights compatible with existing inference engines. Kimi K3 is an open-source project aimed at developers and AI researchers. Kimi K3 is open source under the Apache-2.0 license. It ships for the web, the command line, and API.
Behind Kimi K3 is Unsloth, and it first shipped in 2023. The project is developed in the open on GitHub with 69.1k stars and 1.4k commits in the last 90 days. Among its 3 catalogued features are Quantized GGUF models, Local LLM inference, and model compression.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do