LaunchAI Models
Jun 23, 10:01 PM
Featured
Qwable 27B fine-tune mimics Claude Fable 5 reasoning, runs locally
Qwable 27B is a full fine-tune of Alibaba's Qwen3.6-27B, trained on a Fable 5-style reasoning dataset to replicate structured, deliberate thinking. An abliterated version surgically removes refusal behavior via llama.cpp's cvector-generator; both models run locally on consumer hardware.
·
Jun 23, 10:01 PM