💥 Big news, developers! Cognition and Windsurf have just unleashed SWE-grep and SWE-grep-mini, two power-packed models designed to revolutionize your code search game. These aren’t your average models; they’re built for speed and efficiency, promising to cut your search time by up to 20 times! 🏃♂️💨
Here’s the scoop:
🔹 Blazing Fast & Parallel: SWE-grep and SWE-grep-mini can handle up to 8 parallel tool calls per turn, for up to 4 turns. This means they can find what you need in a flash, even in massive codebases.
🔹 Smart & Precise: Unlike traditional methods, these models use reinforcement learning to prioritize precision. They learn from their mistakes, minimizing context pollution and saving your valuable agent tokens.
🔹 Easy to Use: Windsurf users, rejoice! The new Fast Context subagent is rolling out gradually, and you don’t need to do anything special to use it. Just keep using Windsurf Cascade, and the new feature will kick in when you need to search for code.
🔹 Playground & Cross-Platform Support: Anyone can experiment with these models in the dedicated demo playground. Plus, they support multiple platforms, including Windows, so you’re covered no matter where you code.
Cognition and Windsurf, the powerhouses behind these innovative tools, are determined to keep you “in flow” and boost your productivity. They’re not stopping here, either – plans are already in the works to expand SWE-grep’s deployment across more products and future updates. Stay tuned, developers! Your coding future just got a whole lot faster. 🚀💻
