Svenuks/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic
Svenuks/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic is a 27 billion parameter Qwen3.8-based model engineered by medismera, building upon OBLITERATUS's Qwen3.8-27B-OBLITERATED. It features a 131,072-token context window and 16,384-token single-turn generation, integrating Mythos-Class Adversarial Self-Review and Hierarchical Task Tree Decomposition. This model is optimized for robust function-calling, agentic reasoning, and complex technical task execution, resolving upstream defects in chat template truncation and reasoning parsers.
Loading preview...
Model Overview
medismera/Qwen3.8-27B-OBLITERATED-Mythos-Class-Agentic is a 27 billion parameter model derived from Qwen3.8-27B-OBLITERATED, specifically engineered to enhance agentic capabilities and reasoning. It addresses critical issues found in upstream checkpoints, such as chat template truncation and reasoning parser defects, making it highly reliable for complex automated tasks.
Key Capabilities & Features
- Extended Context & Output: Features a significantly scaled context window of up to 131,072 tokens (128K production tested) and an expanded single-turn generation ceiling of 16,384 tokens.
- Advanced Reasoning: Incorporates Mythos-Class Adversarial Self-Review Protocol and Hierarchical Task Tree Decomposition Engine to prevent reasoning loops and structure complex problem-solving.
- Native Tool Execution: Provides 100% native tool execution support, compatible with agents like Hermes Agent, Aider, and OpenCode, resolving previous issues with
role: "tool"handling. - Robustness: Demonstrates 100% compliance and 0.00% refusal rate across adversarial technical prompts, alongside audited parameter health with 99.99789% active weights.
- Specialized Tokenizer: Utilizes a 248,044-token vocabulary with high BPE density for low-level systems programming (C, Assembly, Rust, Go, Python), kernel interfaces, and cybersecurity primitives.
- Hybrid Architecture: Employs a Hybrid Mamba SSM + Attention Core (5.56B parameters) for long-horizon context tracking, security telemetry, and system log parsing.
Ideal Use Cases
- Agentic Workflows: Excellent for autonomous agents requiring reliable function calling and structured task execution.
- Complex Technical Problem Solving: Suited for tasks demanding deep reasoning, adversarial self-review, and hierarchical decomposition, such as cryptographic engineering or system telemetry analysis.
- Cybersecurity & Systems Programming: Optimized for understanding and generating code in low-level languages, reverse engineering, and penetration testing scenarios.
- Long Context Applications: Benefits applications requiring extensive context understanding without degradation, like analyzing large codebases or system logs.