unsloth/Kimi-Dev-72B
Kimi-Dev-72B is a 72.7 billion parameter open-source coding LLM developed by the Kimi-Dev Team, optimized for software engineering tasks. It achieves a new state-of-the-art performance of 60.4% on SWE-bench Verified among open-source models. The model is specifically trained using large-scale reinforcement learning to autonomously patch real repositories and pass entire test suites, ensuring robust and correct solutions. It is designed for issue resolution and general coding applications.
Loading preview...
Kimi-Dev-72B: A State-of-the-Art Open-Source Coding LLM
Kimi-Dev-72B, developed by the Kimi-Dev Team, is a 72.7 billion parameter open-source coding Large Language Model specifically engineered for software engineering tasks. This model distinguishes itself by achieving a new state-of-the-art performance of 60.4% on SWE-bench Verified, surpassing other open-source models in this benchmark.
Key Capabilities & Differentiators
- Superior Code Issue Resolution: Kimi-Dev-72B is optimized through large-scale reinforcement learning, where it autonomously patches real repositories within Docker environments. It only receives rewards upon successfully passing the entire test suite, ensuring high-quality, correct, and robust solutions aligned with real-world development standards.
- Benchmark Performance: It sets a new benchmark for open-source models on the challenging SWE-bench Verified dataset.
Good For
- Software Development: Ideal for tasks requiring code generation, debugging, and automated issue resolution.
- Research & Development: Provides a strong foundation for further research in code-centric LLMs and reinforcement learning applications in software engineering.
Developers and researchers are encouraged to explore its capabilities for various coding applications.