Evidence item

A Hugging Face model card for a distilled Qwen-based 7B model lists a Q6_K quantization at 6.25 GB, independently corroborating that a distilled Qwen-class model can be compressed to roughly a 6 GB file. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 15450

Source: DeepSeek-R1-Distill-Qwen-7B-GGUF model card · locator {} · excerpt hash sha256:32f24be0a2e818825c3881d955191ae5d5cd5a815668ead2edd5024bdd7016d7

| Q6_K | 6.4 | very good quality |
...
6-bit
Q6_K 6.25 GB