Introducing our new family of models: SWE-1, SWE-1-lite, and SWE-1-mini. Based on internal evals, it has performance nearing that of frontier models from the foundation labs.
Just dropped something big — Windsurf Wave 9 with the new SWE-1 models!
These AI models are built to help with real engineering work — not just writing code. You’ve got three options:
✅ SWE-1: Super powerful for heavy, complex tasks
✅ SWE-1 Lite: Lighter and faster for everyday dev work
✅ SWE-1 Mini: Blazing fast for when speed really matters
They’re all built into the Windsurf Editor, making things like debugging and code generation way easier. It’s like having an extra set of smart hands helping out.
This tool looks super powerful! I’m curious, can these three SWE-1 models be seamlessly switched during use? For example, if I’m working on a complex task with SWE-1 but need to quickly handle a simpler one, can I directly switch to SWE-1 Mini?
Terrible bad models, the worse Windsurf has. Even Claude 3.5 is better. I dont know why they released such an early bad product, I imagine investors.
They asked a reviewed by mail during the weekend and I like windsurf, so I was keen, but the new feedback platform selected for this used a third party system: Gartner Peer Community™ and Gartner Peer Insights™ with their own policies. That pised me off even more that a an early bad release with marketing lies. I dont want to give my data to another third party system with bad rep.
Replies
👋 Hey Hunters!
Just dropped something big — Windsurf Wave 9 with the new SWE-1 models!
These AI models are built to help with real engineering work — not just writing code. You’ve got three options:
✅ SWE-1: Super powerful for heavy, complex tasks
✅ SWE-1 Lite: Lighter and faster for everyday dev work
✅ SWE-1 Mini: Blazing fast for when speed really matters
They’re all built into the Windsurf Editor, making things like debugging and code generation way easier. It’s like having an extra set of smart hands helping out.
Excited to see what you build with it! 🚀🛠️
This tool looks super powerful! I’m curious, can these three SWE-1 models be seamlessly switched during use? For example, if I’m working on a complex task with SWE-1 but need to quickly handle a simpler one, can I directly switch to SWE-1 Mini?
Mukh.1
Impressive launch. Congratulations. We recently launched Mukh.1, a no-code platform for AI workflows and agents. Check it out.
Windsurf is looking more and more powerful—I’m already a loyal user!
Raycast
"Wave 9" isn't the product name — it's SWE-1.
WebCurate.co
I haven't seen any model designed specifically for engineering. Sounds very promising. Is it live yet? I'd like to try it out.
Terrible bad models, the worse Windsurf has. Even Claude 3.5 is better. I dont know why they released such an early bad product, I imagine investors.
They asked a reviewed by mail during the weekend and I like windsurf, so I was keen, but the new feedback platform selected for this used a third party system: Gartner Peer Community™ and Gartner Peer Insights™ with their own policies. That pised me off even more that a an early bad release with marketing lies. I dont want to give my data to another third party system with bad rep.