Model Overview
TeichAI/Qwen3-14B-Gemini-3-Pro-Preview-High-Reasoning-Distill is a 14 billion parameter language model built upon the Qwen3 architecture. It was developed by TeichAI and specifically fine-tuned using a proprietary Gemini 3 Pro Preview dataset, which emphasizes high reasoning effort. This training approach aims to enhance the model's analytical and problem-solving capabilities.
Key Capabilities & Training
- High Reasoning Focus: The model's core differentiator is its training on a dataset designed to impart advanced reasoning skills, derived from a Gemini 3 Pro Preview source.
- Base Architecture: It leverages the
unsloth/Qwen3-14B as its foundational model. - Efficient Training: The model was trained using Unsloth and Huggingface's TRL library, enabling faster training times.
- Context Length: It supports a substantial context window of 32768 tokens.
Ideal Use Cases
This model is particularly well-suited for applications demanding strong logical and analytical processing:
- Coding: Generating, understanding, and debugging code.
- Science: Assisting with scientific research, data analysis, and complex problem-solving in scientific domains.
Related Models
TeichAI has also released smaller, related models in this series, including TeichAI/Qwen3-8B-Gemini-3-Pro-Preview-Distill and TeichAI/Qwen3-4B-Thinking-2507-Gemini-3-Pro-Preview-High-Reasoning-Distill, which share a similar reasoning-focused training methodology.