# LongBench v2

LongBench v2 is an AI benchmark; this page records what it measures and traces its data, grader, and scores to primary sources.

Tests deep reasoning over realistic contexts from 8,000 to two million words using difficult multiple-choice questions.

## Details

- Address: Online, Global
- Website: [longbench2.github.io/](<https://longbench2.github.io/>)
- Tags: Helpful, Science
- Category: Memory
- Status: Active
- Data licence: MIT
- Data access: open
- Grader: exact
- Organization: THUDM
- Released: 2024-12
- Paper: [arxiv.org/abs/2412.15204](<https://arxiv.org/abs/2412.15204>)
- Leaderboard: [longbench2.github.io/](<https://longbench2.github.io/>)
- Repository: [github.com/THUDM/LongBench](<https://github.com/THUDM/LongBench>)
- Task count: 503
- Top-score provenance: unknown
- ABI family: No

## Questions

### Where is LongBench v2?

LongBench v2 is at Online, Global.

### How do I contact LongBench v2?

To contact LongBench v2, visit https://longbench2.github.io/.

## Source

From [Benchmark primary sources](https://aibenchmarkindex.com/about), license CC0-1.0, checked 2026-10-15.

Last updated October 15, 2026.

## Links

- This benchmark on the [web page](<https://aibenchmarkindex.com/global/online/longbench-v2>)
- [Memory in Online](<https://aibenchmarkindex.com/global/online/category/memory.md>)
- [More benchmarks in Online](<https://aibenchmarkindex.com/global/online.md>)

> **Check before you compare.** Benchmark methods and leaderboards change. Follow the cited primary source before using a result in a decision.