Mercury

Mercury Edit 2 is a specialized diffusion-based Large Language Model (dLLM) designed by Inception for highly latency-sensitive coding workflows. Unlike traditional autoregressive LLMs that generate text token by token, Mercury Edit 2 leverages a diffusion framework to produce tokens in parallel, significantly increasing speed and efficiency. This makes it particularly well-suited for real-time coding assistance, such as autocomplete and next-edit suggestions. This model is part of Inception's Mercury family of dLLMs, which aims to provide frontier LLM quality at a fraction of the cost and with superior speed. Mercury Edit 2 is built for developers and teams looking to integrate ultra-responsive AI into their development environments, enabling a more fluid and productive coding experience. Its smaller context window (32K) is optimized for focused coding tasks, ensuring rapid responses without sacrificing quality for its intended use cases. It is OpenAI API compatible, allowing for seamless integration into existing LLM workflows.