reasoning benchmark
LongBench v2
LongBench v2 is a benchmark designed to assess the ability of LLMs to handle long-context problems requiring deep understanding and reasoning across real-world multitasks. It consists of 503 challenging multiple-choice questions with contexts ranging from 8k to 2M words across six major task categories: single-document QA, multi-document QA, long in-context learning, long-dialogue history understanding, code repository understanding, and long structured data understanding.
Updated Aug 7, 2026
Published models17
Registry coverage17
MetricScore
EvidenceC
LongBench v2 leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | AC | 66.3% | 100.0% | 17 | C | |
| 2 | AC | 63.2% | 93.8% | 17 | C | |
| 3 | AC | 62.0% | 87.5% | 17 | C | |
| 4 | NV | 61.9% | 81.3% | 17 | C | |
| 5 | MI | 61.5% | 75.0% | 17 | C | |
| 6 | MA | 61.0% | 68.8% | 17 | C | |
| 7 | MI | 61.0% | 62.5% | 17 | C | |
| 8 | MI | 61.0% | 56.3% | 17 | C | |
| 9 | XI | 60.6% | 50.0% | 17 | C | |
| 10 | AC | 60.6% | 43.8% | 17 | C | |
| 11 | AC | 60.2% | 37.5% | 17 | C | |
| 12 | AC | 59.0% | 31.3% | 17 | C | |
| 13 | AC | 55.2% | 25.0% | 17 | C | |
| 14 | AC | 50.0% | 18.8% | 17 | C | |
| 15 | DE | 48.7% | 12.5% | 17 | C | |
| 16 | AC | 38.7% | 6.3% | 17 | C | |
| 17 | AC | 26.1% | 0.0% | 17 | C |
Score distribution
Top published rows on the benchmark's original scale.
LongBench v2 highlights
The top published results on this benchmark's own scale.
What is LongBench v2?
Definition and scoring fields from the benchmark registry.
LongBench v2 is a benchmark designed to assess the ability of LLMs to handle long-context problems requiring deep understanding and reasoning across real-world multitasks. It consists of 503 challenging multiple-choice questions with contexts ranging from 8k to 2M words across six major task categories: single-document QA, multi-document QA, long in-context learning, long-dialogue history understanding, code repository understanding, and long structured data understanding.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level C.
- Family
- LongBench v2
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- longbench-v2|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about LongBench v2.
Which model scores highest on LongBench v2?
Qwen3.8 Max is currently ranked first with 66.3%.
What does LongBench v2 measure?
LongBench v2 is a benchmark designed to assess the ability of LLMs to handle long-context problems requiring deep understanding and reasoning across real-world multitasks. It consists of 503 challenging multiple-choice questions with contexts ranging from 8k to 2M words across six major task categories: single-document QA, multi-document QA, long in-context learning, long-dialogue history understanding, code repository understanding, and long structured data understanding.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
17 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is marked as eligible for the current LLMBoard capability methodology.