Does anyone use uncensored models purely for coding?

A user on reddit.com inquired whether uncensored models are used purely for coding, noting that some believe they are better as they avoid constant "is this okay" checks. The user tested the HuiHui Qwen 3.8 27b (UD-Q8_K_XL) quant for coding, finding it reasons less but maintains coding quality. After 8 hours of testing, it appears more confident, uses less context, and its reasoning is shorter compared to the Dirk template version of Qwen 3.8 27b.