Kimi-Dev: Agentless Training as Skill Prior for SWE-agents

ICLR 2026 Conference Submission17230 Authors

Keywords: coder LLM, Agentless, SWE-Agent, Reinforcement Learning

TL;DR: We present Kimi-Dev that obtains 60.4% pass rate on SWE-bench Verified with Agentless training, and demonstrate it provides strong skill priors that enable efficient and effective SWE-Agent adaptation.

Abstract: Large Language Models (LLMs) are increasingly applied to software engineering (SWE), with SWE-bench as a key benchmark. Solutions are split into SWE-Agent frameworks with multi-turn interactions and workflow-based Agentless methods with single-turn verifiable steps. We argue these paradigms are not mutually exclusive: reasoning-intensive Agentless training induces skill priors, including localization, code edit, and self-reflection that enable efficient and effective SWE-Agent adaptation. In this work, we first curate the Agentless training recipe and present Kimi-Dev, an open-source SWE LLM achieving 60.4\% on SWE-bench Verified, the best among workflow approaches. With additional SFT adaptation on 5k publicly-available trajectories, Kimi-Dev powers SWE-Agents to 48.6\% pass@1, on par with that of Claude 3.5 Sonnet (241022 version). These results show that structured skill priors from Agentless training can bridge workflow and agentic frameworks for transferable coding agents.

Primary Area: foundation or frontier models, including LLMs

Submission Number: 17230