@misc{indiciae2935d745408f,
  title = {Capability-Routed Visual Retrieval and Evidence Threading for Long-Context Document Question Answering},
  author = {Amirul Rahman and Aisha Karim and Kenji Nakamura and Yi-Fan Ng},
  year = {2026},
  url = {https://arxiv.org/abs/2609.13268},
  note = {Source identifier: 2609.13268}
}