Pq234/Darwin-28B-Opus
Pq234/Darwin-28B-Opus is a 27.6 billion parameter language model from the Darwin series, built on the Qwen3.6 generation backbone. It is an evolutionary merge combining Qwen3.6-27B's strong bilingual reasoning with Claude Opus 4-style chain-of-thought distilled behavior. This model excels at graduate-level STEM reasoning, achieving 88.89% on the GPQA Diamond benchmark, and is optimized for complex multi-step reasoning tasks.
Loading preview...
Darwin-28B-Opus: Qwen3.6-Generation Reasoning Model
Darwin-28B-Opus is a 27.6 billion parameter model, part of the Darwin series, leveraging the Qwen3.6 generation architecture. It is the result of a Darwin V7 evolutionary merge, combining the robust bilingual reasoning capabilities of Qwen3.6-27B with a Claude Opus 4-style chain-of-thought reasoning distilled behavior. This model is specifically designed for advanced reasoning tasks.
Key Capabilities
- Exceptional Reasoning Performance: Achieves 88.89% on the GPQA Diamond graduate-level reasoning benchmark using a 3-stage adaptive evaluation protocol, outperforming its 36B MoE sibling and previous Darwin-27B-Opus.
- Evolutionary Merge: Created by merging a Qwen3.6-27B 'father' model with a Claude Opus reasoning-distilled 'mother' variant, inheriting the Qwen3.6 architecture and Opus reasoning style.
- Hybrid Attention: Utilizes a hybrid linear/full attention mechanism, inherited from its Qwen3.6-27B base.
- Bilingual Reasoning: Offers strong reasoning capabilities in English, with secondary support for Korean, Chinese, and Japanese.
Recommended Use Cases
- Graduate-level STEM reasoning: Ideal for tasks like GPQA and science qualifying exams.
- Mathematical problem-solving: Suitable for complex problems such as MATH and AIME-style challenges.
- Complex multi-step chain-of-thought tasks: Designed to handle intricate reasoning processes.
- Code generation and debugging: Applicable for tasks like HumanEval and MBPP.