DDSWE
A benchmark used by DeepSeek to evaluate model programming capability. Its tasks include building complete projects from scratch, fixing code defects, and implementing specified functionality, requiring models to output runnable code in realistic engineering environments.
Timeline (last 90 days)
DeepSeek V4.1 Flash L017 reaches a 77% pass rate on the DDSWE coding benchmark, at an average cost of only 1.3 yuan per task.
DDSWE is a benchmark DeepSeek uses to evaluate models' coding ability, including tasks such as building a complete project from scratch, fixing code defects and implementing specified features, and requires the model to output runnable code.
All Facts (2)
DeepSeek V4.1 Flash L017 reaches a 77% pass rate on the DDSWE coding benchmark, at an average cost of only 1.3 yuan per task.
50%UnverifiedDDSWE is a benchmark DeepSeek uses to evaluate models' coding ability, including tasks such as building a complete project from scratch, fixing code defects and implementing specified features, and requires the model to output runnable code.
50%