A realistic long-context understanding and reasoning benchmark with 503 multiple-choice questions spanning contexts from 8,000 to 2 million words. Tasks include single- and multi-document QA, long conversations, code repositories, and structured data.
Browse the latest scores, model modes, release dates, and parameter sizes for LongBench v2.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 Opus 4.5 Standard Mode | 64.40 | 2025-11-25 | Unknown | Closed | |
 Qwen3.5-397B-A17B Standard Mode | 63.20 | 2026-02-16 | 39.7B | Free Commercial | |
Qwen 3.6 Plus Preview Standard Mode | 62.00 | 2026-03-31 | Unknown | Closed | |
4 |  Nemotron 3 Ultra Standard Mode | 61.90 | 2026-06-04 | 550B | Free Commercial |
5 |  Kimi K2.5 Standard Mode | 61.00 | 2026-01-27 | 1000B | Free Commercial |
6 |  GLM-5 Standard Mode | 60.80 | 2026-02-11 | 744B | Free Commercial |
7 | Qwen3.5-27B Standard Mode | 60.60 | 2026-02-25 | 27B | Free Commercial |
8 | Qwen3.5-122B-A10B Standard Mode | 60.20 | 2026-02-25 | 122B | Free Commercial |
9 | Qwen3.5-35B-A3B Standard Mode | 59.00 | 2026-02-25 | 35B | Free Commercial |
10 |  DeepSeek-V4-Pro-Base Standard Mode | 51.50 | 2026-04-24 | 1600B | Free Commercial |
11 | DeepSeek-V4-Flash-Base Standard Mode | 44.70 | 2026-04-24 | 284B | Free Commercial |
