IQuest-Coder-V1 Technical Report
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917350255624192 |
|---|---|
| author | Yang, Jian Zhang, Wei Guo, Shawn Ye, Zhengmao Jing, Lin Liu, Shark Li, Yizhi Wu, Jiajun Liu, Cening Ma, X. Song, Yuyang Wu, Siwei Li, Yuwen Liao, L. Zheng, T. Huang, Ziling Huang, Zelong Liu, Che Xing, Yan Li, Renyuan Cai, Qingsong Yan, Hanxu Wang, Siyue Li, Shikai Liu, Jason Klein Huang, An Kang, Yongsheng Zhang, Jinxing Hao, Chuan Wang, Haowen Gu, Weicheng Tao, Ran Tang, Mingjie Wu, Peihao Wang, Jianzhou Liu, Xianglong Lv, Weifeng Dai, Bryan |
| author_facet | Yang, Jian Zhang, Wei Guo, Shawn Ye, Zhengmao Jing, Lin Liu, Shark Li, Yizhi Wu, Jiajun Liu, Cening Ma, X. Song, Yuyang Wu, Siwei Li, Yuwen Liao, L. Zheng, T. Huang, Ziling Huang, Zelong Liu, Che Xing, Yan Li, Renyuan Cai, Qingsong Yan, Hanxu Wang, Siyue Li, Shikai Liu, Jason Klein Huang, An Kang, Yongsheng Zhang, Jinxing Hao, Chuan Wang, Haowen Gu, Weicheng Tao, Ran Tang, Mingjie Wu, Peihao Wang, Jianzhou Liu, Xianglong Lv, Weifeng Dai, Bryan |
| contents | In this report, we introduce the IQuest-Coder-V1 series-(7B/14B/40B/40B-Loop), a new family of code large language models (LLMs). Moving beyond static code representations, we propose the code-flow multi-stage training paradigm, which captures the dynamic evolution of software logic through different phases of the pipeline. Our models are developed through the evolutionary pipeline, starting with the initial pre-training consisting of code facts, repository, and completion data. Following that, we implement a specialized mid-training stage that integrates reasoning and agentic trajectories in 32k-context and repository-scale in 128k-context to forge deep logical foundations. The models are then finalized with post-training of specialized coding capabilities, which is bifurcated into two specialized paths: the thinking path (utilizing reasoning-driven RL) and the instruct path (optimized for general assistance). IQuest-Coder-V1 achieves state-of-the-art performance among competitive models across critical dimensions of code intelligence: agentic software engineering, competitive programming, and complex tool use. To address deployment constraints, the IQuest-Coder-V1-Loop variant introduces a recurrent mechanism designed to optimize the trade-off between model capacity and deployment footprint, offering an architecturally enhanced path for efficacy-efficiency trade-off. We believe the release of the IQuest-Coder-V1 series, including the complete white-box chain of checkpoints from pre-training bases to the final thinking and instruction models, will advance research in autonomous code intelligence and real-world agentic systems. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_16733 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | IQuest-Coder-V1 Technical Report Yang, Jian Zhang, Wei Guo, Shawn Ye, Zhengmao Jing, Lin Liu, Shark Li, Yizhi Wu, Jiajun Liu, Cening Ma, X. Song, Yuyang Wu, Siwei Li, Yuwen Liao, L. Zheng, T. Huang, Ziling Huang, Zelong Liu, Che Xing, Yan Li, Renyuan Cai, Qingsong Yan, Hanxu Wang, Siyue Li, Shikai Liu, Jason Klein Huang, An Kang, Yongsheng Zhang, Jinxing Hao, Chuan Wang, Haowen Gu, Weicheng Tao, Ran Tang, Mingjie Wu, Peihao Wang, Jianzhou Liu, Xianglong Lv, Weifeng Dai, Bryan Artificial Intelligence Computation and Language Software Engineering In this report, we introduce the IQuest-Coder-V1 series-(7B/14B/40B/40B-Loop), a new family of code large language models (LLMs). Moving beyond static code representations, we propose the code-flow multi-stage training paradigm, which captures the dynamic evolution of software logic through different phases of the pipeline. Our models are developed through the evolutionary pipeline, starting with the initial pre-training consisting of code facts, repository, and completion data. Following that, we implement a specialized mid-training stage that integrates reasoning and agentic trajectories in 32k-context and repository-scale in 128k-context to forge deep logical foundations. The models are then finalized with post-training of specialized coding capabilities, which is bifurcated into two specialized paths: the thinking path (utilizing reasoning-driven RL) and the instruct path (optimized for general assistance). IQuest-Coder-V1 achieves state-of-the-art performance among competitive models across critical dimensions of code intelligence: agentic software engineering, competitive programming, and complex tool use. To address deployment constraints, the IQuest-Coder-V1-Loop variant introduces a recurrent mechanism designed to optimize the trade-off between model capacity and deployment footprint, offering an architecturally enhanced path for efficacy-efficiency trade-off. We believe the release of the IQuest-Coder-V1 series, including the complete white-box chain of checkpoints from pre-training bases to the final thinking and instruction models, will advance research in autonomous code intelligence and real-world agentic systems. |
| title | IQuest-Coder-V1 Technical Report |
| topic | Artificial Intelligence Computation and Language Software Engineering |
| url | https://arxiv.org/abs/2603.16733 |