CloudGoat/Orythos-9B
CloudGoat/Orythos-9B is a 9 billion parameter language model with a 32768 token context length, developed by CloudGoat. It is a merge of empero-ai/Qwythos-9B and ornith-ai/Ornith-1.5-9B, built upon Qwen/Qwen3.5-9B using the DARE-TIES method. This model is specifically designed to combine Qwythos's front-end design and reasoning capabilities with Ornith's agent-like functionality, excelling particularly in coding tasks.
Loading preview...
Orythos-9B: Merged Model for Enhanced Coding and Agentic Capabilities
Orythos-9B is a 9 billion parameter model created by CloudGoat, merging the strengths of two distinct models: empero-ai/Qwythos-9B and ornith-ai/Ornith-1.5-9B. Built on the Qwen/Qwen3.5-9B base, this model leverages the DARE-TIES merge method, with specific layer-wise parameter weighting determined by CloudGoat's RoMM router.
Key Capabilities & Differentiators
- Enhanced Agent-like Functionality: Integrates Ornith's agent-like behaviors with Qwythos's reasoning and front-end design specialization.
- Strong Coding Performance: Achieves a HumanEval pass@1 score of 89.0%, ranking second among community models and solving approximately 146 out of 164 problems. This significantly surpasses the performance of its Qwythos-9B component (79.9%).
- Advanced Merge Technique: Utilizes the DARE-TIES method, a sophisticated merging technique, to combine model capabilities effectively.
Ideal Use Cases
- Code Generation and Completion: Its high HumanEval score indicates strong proficiency in programming tasks.
- Agentic Applications: Suitable for scenarios requiring models to exhibit more autonomous, agent-like behaviors.
- Complex Reasoning with Front-end Design: Benefits from Qwythos's original strengths in reasoning and front-end design, now augmented with coding and agentic features.