Unsloth released Laguna-S-2.1 in GGUF format to enable efficient local inference on consumer hardware. This quantization update allows users to run the model with reduced memory overhead and improved sampling speeds across various quantized bit depths.