DedeProGames/NTX-2.1-Pro
DedeProGames/NTX-2.1-Pro is a 14 billion parameter model, distilled from GPT-5.5 reasoning traces, designed for enhanced instruction following and logical reasoning. This compact model excels in code generation and general instruction tuning, offering strong performance for its size. With a 32768 token context length, it provides efficient processing for complex tasks. Its primary strength lies in improving reasoning and coding capabilities while maintaining a small operational footprint.
Loading preview...
NTX-2.1-Pro: Compact Reasoning and Code Generation
NTX-2.1-Pro is a 14 billion parameter model developed by DedeProGames, specifically engineered for improved instruction tuning, logical reasoning, and code generation. It stands out due to its distillation from GPT-5.5 reasoning traces, which imbues it with advanced capabilities typically found in much larger models, yet it maintains a compact size for efficiency.
Key Capabilities
- Enhanced Reasoning: Benefits from GPT-5.5 distillation for superior logical reasoning and instruction following.
- Strong Code Generation: Delivers competitive performance in coding tasks for its parameter count.
- Efficient Operation: Its 14B parameters make it more efficient to deploy and run compared to larger alternatives.
- Competitive Benchmarking: Achieves a GPQA score of 67.4, outperforming Qwen3-14B (Thinking) at 64.0, indicating strong performance in complex question answering.
When to Use NTX-2.1-Pro
This model is ideal for developers and applications requiring:
- Resource-constrained environments where larger models are impractical.
- Tasks demanding high logical reasoning and precise instruction adherence.
- Code generation applications where accuracy and efficiency are critical.
- General instruction-following scenarios that benefit from advanced distillation techniques.