Show HN: Abliterated LLM provider for cyber tasks

Category: ai-ml

Tags: llm, abliterated-models, api-wrapper, openai-compatible, uncensored

Score: 3.7/10 (Innovation: 4, Technical: 3, Documentation: 3, Utility: 4)

Refuseless is a hosted API service providing access to abliterated LLM models (GLM, DeepSeek) with tool-calling and reasoning support, integrated via an OpenCode config snippet. It targets users who need uncensored model outputs for 'cyber tasks', but the project itself is essentially a thin commercial wrapper with minimal open-source substance.

Target audience: AI developers and security researchers seeking uncensored LLM endpoints

Repository: https://refuseless.com

View on Hacker News