Modelpublic

OpenHands/CodeScout-4B

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face

1likes562downloads

Model Card

CodeScout-4B

Best efficiency–performance trade-off — outperforms 8× larger Qwen3-32B across all benchmarks.

CodeScout-4B is part of the CodeScout family of open-source RL-trained code search agents. CodeScout models achieve state-of-the-art repository-level code localization using nothing more than a standard Unix terminal — no static analysis, no repository graphs, no language-specific tooling.

Key Highlights

—Consistently outperforms 8× larger Qwen3-32B on all benchmarks
—Surpasses RepoNavigator-14B by 2–10% in file F1 and 8–11% in function F1
—Exceeds GPT-5 with RepoNavigator by 9% in file F1 and 5% in function F1 on SWE-Bench Verified
—Best efficiency–performance trade-off in the CodeScout family

Results

Performance on SWE-Bench code localization (instance-averaged F1 scores):

Benchmark	CodeScout-1.7B	CodeScout-4B	CodeScout-14B
SWE-Bench Verified — File F1	55.46	68.52	68.57
SWE-Bench Verified — Func F1	28.22	36.78	40.32
SWE-Bench Pro — File F1	40.96	51.77	53.63
SWE-Bench Pro — Func F1	18.24	29.03	28.74
SWE-Bench Lite — File F1	56.57	67.03	71.84
SWE-Bench Lite — Func F1	27.07	39.87	44.43

Code localization performance on SWE-Bench Verified. CodeScout (⭐) achieves superior or competitive results over larger open-source LLMs and narrows the gap with closed-source frontier models.

Training

CodeScout-4B is trained directly from Qwen3-4B-Instruct-2507 using GSPO reinforcement learning.

—Training data: 1,600 instances from SWE-Smith (39K filtered, 128 repos)
—RL steps: 200
—Batch size: 8, with 8 rollouts per instance
—Max context length: 40K tokens
—Max turns per episode: 6
—Reward: Multi-level F1 (file + module + function)
—Hardware: 8×H100 GPUs
—Learning rate: 1e-6 (constant)

How It Works

CodeScout uses the OpenHands-Bash scaffold — an agent equipped with only a Terminal tool (supporting standard Unix commands like rg, find, grep, ls) and a LocalizationFinish tool for structured output submission. The agent iteratively navigates the repository to identify relevant files, classes, and functions related to a given issue.

The model is trained with GSPO (Group Sequence Policy Optimization) using multi-level F1 rewards at the file, module, and function level.

Intended Use

CodeScout-4B is designed for repository-level code localization: given a GitHub issue description and a code repository, it identifies the relevant files, classes, and functions that need to be modified. It is intended to be used as a localization subagent within larger coding agent pipelines.

Limitations

—Trained and evaluated exclusively on Python repositories
—Designed for code localization, not code editing or issue resolution
—Performance may vary on repositories significantly different from the training distribution
—Requires the OpenHands-Bash scaffold for optimal performance

Citation

bibtex

@misc{sutawika2026codescouteffectiverecipereinforcement,
      title={CodeScout: An Effective Recipe for Reinforcement Learning of Code Search Agents}, 
      author={Lintang Sutawika and Aditya Bharat Soni and Bharath Sriraam R R and Apurva Gandhi and Taha Yassine and Sanidhya Vijayvargiya and Yuchen Li and Xuhui Zhou and Yilin Zhang and Leander Melroy Maben and Graham Neubig},
      year={2026},
      eprint={2603.17829},
      archivePrefix={arXiv},
      primaryClass={cs.SE},
      url={https://arxiv.org/abs/2603.17829}, 
}