COMPINT Benchmark

What is COMPINT?

COMPINT measures how well context compactors preserve user-issued session constraints across chat, agentic, and research scenarios.

Released

2026-07-31

Evaluates

General AI, Language & Knowledge, Long Context

Openness

unknown

Importer

Claire Radar

Review status

ai-reviewed

Reported scores

0

Source provenance

• Original evidence https://arxiv.org/abs/2608.11242

COMPINT paper and code

• Paper https://arxiv.org/abs/2608.11242
• Code https://github.com/ZhiqiEliWang/compaction-integrity

Which sources cite COMPINT?

• COMPINT

Related General AI, Language & Knowledge, Long Context benchmarks

• $\tau^\tau$-Bench: An Environment for End-To-End, Realistic Agent Construction
• 3DCodeBench
• 5-Dialects-BN
• A Large-Scale Dataset and Benchmark
• A3S-Bench
• AARRI-Bench
• ABC-Bench
• ActBench