openchat/openchat-3.5-0106

OpenChat/openchat-3.5-0106 is a 7 billion parameter language model developed by OpenChat, optimized for chat, coding, and mathematical reasoning. This iteration significantly improves coding performance, achieving 71.3 on HumanEval and outperforming ChatGPT (March) and Grok-1 on various benchmarks. It features two distinct modes for generalist/coding tasks and mathematical problem-solving, making it suitable for diverse applications requiring strong analytical capabilities.