---
license: mit
pretty_name: InteractBench
language:
- en
tags:
- code
- benchmark
- competitive-programming
- interactive
- llm-evaluation
---
# InteractBench Dataset

Dataset artifacts for **InteractBench: Benchmarking LLMs on Competitive Programming under Unrevealed Information** (ICML 2026).

[Project page](https://www.interactbench.com) | [Paper (OpenReview)](https://openreview.net/forum?id=Y4T4w0Tj0l) | [Code](https://github.com/kmsgk0/InteractBench)

InteractBench is a benchmark of 322 interactive competitive-programming problems curated from Codeforces, AtCoder, IOI, and ICPC. Each problem ships with an executable local interactor, so evaluation runs fully offline without external judge submission.

Files:
- `problems.jsonl`: 298 standard interactive problems.
- `ioi.jsonl`: 24 IOI-style problems.
- `test_cases/*.tar.zst`: materialized problem test cases split into shards.
- `checksums.sha256`: SHA-256 checksums for released files.

Use:

```bash
sha256sum -c checksums.sha256
mkdir -p data/problems
for f in test_cases/*.tar.zst; do
  tar --use-compress-program=unzstd -xf "$f" -C data/problems
done
python scripts/import_from_jsonl.py --type standard --input problems.jsonl --output-dir data/problems
python scripts/import_from_jsonl.py --type ioi --input ioi.jsonl --output-dir data/problems
```

License: MIT.

## Citation

```bibtex
@inproceedings{li2026interactbench,
  title     = {InteractBench: Benchmarking {LLM}s on Competitive Programming under Unrevealed Information},
  author    = {Jiaze Li and Aocheng Shen and Bing Liu and Boyu Zhang and Xiaoxuan Fan and Qiankun Zhang and Xianjun Deng},
  booktitle = {Forty-third International Conference on Machine Learning},
  year      = {2026},
  url       = {https://openreview.net/forum?id=Y4T4w0Tj0l}
}
```