Megadose The AI curve, compressed.

Termination Lag: Coding Agents Solve Early but Stop Late

· HF Daily Papers ·
In 38–82% of successful runs, coding agents produced a correct solution on their first source-code edit.

Researchers tested every distinct repository patch across five models, two agent frameworks and three SWE-bench datasets. On SWE-bench Verified and Multilingual, steps after the first correct patch accounted for a median 35–37% of successful runs for Qwen3-Coder and 54–60% for Laguna-XS.2. Most of that additional work went into verification and repository exploration. In 1–8% of cases, continued activity undid an earlier correct patch, leaving the final solution incorrect.

HF Daily Papers' note

score 5

Categories: Research