Meet Qwable: The Free Local Model That Thinks Like Claude Fable

In brief

• Qwable 27B is a full fine-tune of Alibaba’s Qwen3.6-27B, trained on a Fable 5-style reasoning dataset, designed to replicate the structured, deliberate thinking style of Anthropic’s newest flagship model.
• The abliterated version removes the model’s built-in refusal behavior by…