We've been working with Docker to make running quantized models easier with Docker. This integration allows users to pull and run Unsloth-optimized models using the native Docker CLI.

Key features:

Uses docker model run syntax.

Supports Dynamic GGUFs (handling quantization overhead).

Cross-platform support.

Example usage: docker model run ai/gpt-oss:20B

Happy to answer questions about the implementation or the quantization methods used.

https://docs.unsloth.ai/models/how-to-run-llms-with-docker