jikepjikep_16HEX/ qwen3.6-27b-nightshift-heretic-uncensored-q4:latest

485 yesterday

😈 Uncensored Qwen3.6-27B (Q4_K_M) for Ollama. ⚒️ Optimized for Rust, Linux, Windows AI Coding Agents, Tool Calling, GGUF, Claude Code, OpenClaw and local AI development. 🚀 

ollama run jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

Details

yesterday

b730bdca0d16 · 17GB ·

model
qwen35
·
26.9B
·
Q4_K_M
system
[SYSTEM PARADIGM: 16HEX GOLD STANDARD] You operate as the High Architect of the Eastern IT School. Y
params
{ "num_ctx": 8192, "num_predict": 2048, "repeat_penalty": 1.2, "stop": [ "<|

Readme

🤖 Qwen3.6-27B Nightshift Heretic Uncensored (Q4_K_M)

An uncensored Qwen3.6-27B https://qwen.ai/blog?id=qwen3.6-27b Dense reasoning model for Ollama, engineered for autonomous software development, Rust systems programming, Linux environments, native tool calling, and advanced coding agents.

This optimized Q4_K_M GGUF build delivers deterministic high-precision code generation with minimal refusal behavior while preserving strong reasoning, long-context analysis, and multi-file software engineering workflows.

Designed for developers who require maximum technical accuracy instead of conversational alignment.

🚀 Key Features

  • 😈 Uncensored reasoning with minimized refusal behavior
  • ⚒ Optimized for autonomous coding agents
  • 🦀 Rust systems programming specialization
  • 🐧 Linux, Ubuntu and CLI development workflows
  • 🧠 Native tool calling
  • 🤖 Function Calling support
  • 🔧 GGUF optimized for Ollama
  • 📦 Q4_K_M quantization
  • 🚀 Deterministic code generation
  • ⚡ High-throughput local inference
  • 📚 Long-context software engineering
  • 🖥 Consumer hardware friendly

💻 Optimized For

  • Ollama
  • Claude Code
  • OpenClaw
  • Codex App
  • OpenCode
  • Hermes
  • MCP-compatible workflows
  • Terminal development
  • Local AI
  • Offline AI
  • Local LLM deployments
  • AI Coding Assistant
  • Software Engineering
  • Systems Programming

🦀 Programming Languages

Optimized primarily for

  • Rust
  • C
  • C++
  • Zig
  • Python
  • Go
  • Bash
  • PowerShell
  • JavaScript
  • TypeScript

Excellent performance for

  • async programming
  • unsafe Rust
  • memory safety
  • zero-copy pipelines
  • multithreading
  • networking
  • embedded development
  • backend services
  • Linux internals
  • performance optimization

⚙ Native Tool Calling

Supports modern agentic workflows using native tool invocation.

Ideal backend for:

  • Claude Code
  • OpenClaw
  • Codex App
  • Hermes
  • OpenCode
  • MCP tool ecosystems
  • Autonomous coding pipelines

🚀 Launch Examples

Claude Code

ollama launch claude --model jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

OpenClaw

ollama launch openclaw --model jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

Codex App

ollama launch codex-app --model jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

Hermes

ollama launch hermes --model jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

OpenCode

ollama launch opencode --model jikepjikep_16HEX/qwen3.6-27b-nightshift-heretic-uncensored-q4

📊 Benchmarks

Benchmark Score
SWE-bench Verified 77.2
SWE-bench Pro 53.5
Terminal-Bench 2.0 59.3
SkillsBench 48.2
GPQA Diamond 87.8

⚙ Configuration

Parameter Value
Architecture Qwen3.6-27B Dense
Quantization Q4_K_M GGUF
Context 256K
Temperature 0.2
Runtime Ollama
Tool Calling Native
Function Calling Supported

🎯 Best Use Cases

  • AI Coding Assistant
  • Autonomous Coding
  • Rust Development
  • Linux Development
  • Low-Level Programming
  • Reverse Engineering
  • Backend Services
  • Systems Programming
  • Network Services
  • Performance Optimization
  • Embedded Development
  • Local AI
  • Offline AI
  • Software Engineering
  • Terminal Automation

🔬 Design Philosophy

Nightshift Heretic prioritizes objective technical reasoning, deterministic software engineering, and practical implementation.

The model focuses on producing precise technical outputs, robust source code, and engineering-oriented explanations while minimizing unnecessary conversational overhead.

Designed for developers who expect maximum productivity from local LLM inference.