Windsurf Releases SWE-1 AI Model Family for Full Software Engineering Lifecycle

Windsurf has launched SWE-1, its first proprietary family of AI models focused on accelerating the full software engineering lifecycle. Unlike existing models that centre on code generation, SWE-1 is designed to support workflows across the entire development process, including incomplete projects and long-running tasks.

The SWE-1 model family includes three purpose-built models for different stages of software development. SWE-1 is the full-scale version used for advanced reasoning and tool use. 

SWE-1-lite, a smaller variant that replaces Cascade Base, is now available to all users. SWE-1-mini is a lightweight model integrated into the Windsurf Tab, offering fast, passive code prediction.

Each of these models is now available to both free and paid users of the Windsurf platform.

“Writing code is only a fraction of what engineers do,” said Varun Mohan, CEO and co-founder of Windsurf. “To truly accelerate software development by 99%, we had to move beyond ‘coding-capable’ models and build software engineering-native models. SWE-1 is our first step in that direction.”

The SWE-1 models operate based on a design principle Windsurf calls flow awareness, the idea that humans and AI share a single development timeline. This allows models to work in step with developers as they write, review, and debug code in the Windsurf Editor.

“Flow awareness lets us see exactly where models succeed or fail, down to the individual decision point,” said Anshul Ramachandran, a member of Windsurf’s founding team. “That feedback loop is our competitive edge.”

According to Windsurf, real-time feedback from users is central to improving the models’ effectiveness. SWE-1 is considered a foundational step in building this shared timeline and may serve as a proof of concept for future end-to-end software engineering automation.

While SWE-1 marks the company’s first entry into frontier model development, Windsurf plans to continue expanding its AI research and development. The company stated that future efforts will focus on scaling the SWE model family and growing the machine learning research team.Engineers interested in testing the new models can access them through the Windsurf Editor at windsurf.com.