ImposterOnline/Pyrex-8B-Instruct-Uncensored

TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Oct 1, 2026License:mitArchitecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Pyrex-8B-Instruct-Uncensored by ImposterOnline is a 7.6 billion parameter uncensored coding model built on the Pyrex 8B Transformer architecture, featuring RoPE, SwiGLU, RMSNorm, and GQA. It is specifically designed for advanced software engineering tasks, multi-turn debugging, and agentic function calling, achieving an 87.20% pass@1 on HumanEval. The model supports a native context length of 32,768 tokens and is optimized for high-level coding applications.

Loading preview...

Pyrex-8B-Instruct-Uncensored: High-Performance Uncensored Coding Model

Pyrex-8B-Instruct-Uncensored is a 7.6 billion parameter model developed by ImposterOnline, specifically engineered for advanced software development tasks. Built on the Pyrex 8B Transformer architecture, it incorporates modern features like RoPE, SwiGLU, RMSNorm, and GQA, making it a robust solution for complex coding challenges. This model is notable for its uncensored nature, providing direct technical responses without built-in safety filters.

Key Capabilities

  • Exceptional Coding Performance: Achieves an impressive 87.20% pass@1 on HumanEval, demonstrating strong code generation abilities.
  • Multi-turn Debugging: Designed to handle complex debugging scenarios across multiple interactions.
  • Agentic Function Calling: Optimized for integrating with agent-based systems for automated task execution.
  • Software Engineering Focus: Excels in algorithm design, system architecture, and high-level software development.
  • Uncensored Output: Provides direct, unfiltered technical responses, ideal for specialized development environments.

Good for

  • Software Engineers: Ideal for generating, debugging, and refining code in various programming languages.
  • Automated Development Agents: Suitable for integration into AI agents requiring precise function calling and code manipulation.
  • Research & Development: Useful for exploring advanced coding paradigms and experimenting with unconstrained AI outputs.
  • Complex Problem Solving: Effective for tackling intricate algorithmic challenges and system design problems.