What changed
Windsurf released SWE-1, SWE-1-lite, and SWE-1-mini for its editor, describing training around incomplete states, tool use, terminals, user interventions, and a shared software-engineering timeline rather than code completion alone.
Why it matters
A developer-tool company used product interaction data and workflow context to specialize models for the complete engineering loop, a defensible advantage beyond access to a generic foundation model.
What to do
Evaluate SWE-1 on ongoing, partly completed work with human course corrections, not only clean benchmark issues; measure retained changes and rework.
Caveat
The models were tied to Windsurf surfaces, not released as open weights. Benchmarks and production experiments were designed and reported by Windsurf, which acknowledged task-specific overfitting.