Gen-Verse/ReasonFlux-Coder-14B is a 14 billion parameter language model developed by Gen-Verse, trained with the CURE algorithm to co-evolve coding and unit test generation abilities. This model excels at code-related tasks, outperforming similarly sized Qwen Coders, DeepSeek Coders, and Seed-Coders. With a 32768 token context length, it is optimized for integration into common test-time scaling and agentic coding pipelines.
No reviews yet. Be the first to review!