Paper Detail

Termination Lag: Coding Agents Solve Early but Stop Late

Musa Cim, Burak Topcu, Mahmut Taylan Kandemir

huggingface Score 7.0

Published 2026-11-11 · First seen 2026-11-17

General AI

Abstract

Coding agents can solve a task well before they stop working on it. We characterize termination lag--the number of steps between an agent's first correct patch and the end of its run--across five models, two agent frameworks, and three SWE-bench datasets by recording and testing every distinct repository patch generated during execution. Our analysis reveals that 38--82% of successful runs achieve a correct solution on their initial source-code edit. However, agents continue to expend substantial effort after correct patches, with post-solution steps accounting for a median of 35--37% for Qwen3-Coder and 54--60% for Laguna-XS.2 in successful runs on SWE-bench Verified and Multilingual. Agents spend most of these additional steps verifying their work and exploring the repository. In a small fraction of cases (1--8%), this continued activity leads agents to undo their earlier correct patch, leaving the final solution incorrect.

Workflow Status

Review status
pending
Role
unreviewed
Read priority
later
Vote
Not set.
Saved
no
Collections
Not filed yet.
Next action
Not filled yet.

Reading Brief

No structured notes yet. Add `summary_sections`, `why_relevant`, `claim_impact`, or `next_action` in `papers.jsonl` to enrich this view.

Why It Surfaced

No ranking explanation is available yet.

Tags

No tags.

BibTeX

@misc{cim2026termination,
  title = {Termination Lag: Coding Agents Solve Early but Stop Late},
  author = {Musa Cim and Burak Topcu and Mahmut Taylan Kandemir},
  year = {2026},
  abstract = {Coding agents can solve a task well before they stop working on it. We characterize termination lag--the number of steps between an agent's first correct patch and the end of its run--across five models, two agent frameworks, and three SWE-bench datasets by recording and testing every distinct repository patch generated during execution. Our analysis reveals that 38--82\% of successful runs achieve a correct solution on their initial source-code edit. However, agents continue to expend substantia},
  url = {https://huggingface.co/papers/2611.14960},
  keywords = {coding agents, termination lag, agent frameworks, SWE-bench, repository patch, source-code edit, Qwen3-Coder, Laguna-XS.2, SWE-bench Verified, SWE-bench Multilingual, code available, huggingface daily},
  eprint = {2611.14960},
  archiveprefix = {arXiv},
}

Metadata

{}