long context benchmark
LongBench v2 Leaderboard
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 17, 2026
Models17
Model coverage17
MetricScore
EvidenceC
LongBench v2 Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAC | Score66.3% | Percentile100.0% | Participants17 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score63.2% | Percentile93.8% | Participants17 | EvidenceC | Evaluated |
| Rank03 | ModelAC | Score62.0% | Percentile87.5% | Participants17 | EvidenceC | Evaluated |
| Rank04 | ModelNV | Score61.9% | Percentile81.3% | Participants17 | EvidenceC | Evaluated |
| Rank05 | ModelMI | Score61.5% | Percentile75.0% | Participants17 | EvidenceC | Evaluated |
| Rank06 | ModelMA | Score61.0% | Percentile68.8% | Participants17 | EvidenceC | Evaluated |
| Rank07 | ModelMI | Score61.0% | Percentile62.5% | Participants17 | EvidenceC | Evaluated |
| Rank08 | ModelMI | Score61.0% | Percentile56.3% | Participants17 | EvidenceC | Evaluated |
| Rank09 | ModelXI | Score60.6% | Percentile50.0% | Participants17 | EvidenceC | Evaluated |
| Rank10 | ModelAC | Score60.6% | Percentile43.8% | Participants17 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score60.2% | Percentile37.5% | Participants17 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score59.0% | Percentile31.3% | Participants17 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score55.2% | Percentile25.0% | Participants17 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score50.0% | Percentile18.8% | Participants17 | EvidenceC | Evaluated |
| Rank15 | ModelDE | Score48.7% | Percentile12.5% | Participants17 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score38.7% | Percentile6.3% | Participants17 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score26.1% | Percentile0.0% | Participants17 | EvidenceC | Evaluated |
LongBench v2 Highlights
The leading models and scores on this benchmark.
LongBench v2 Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for LongBench v2
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
Ranking basisThis longbench v2 AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01ACQwen3.8 MaxAlibaba Cloud / Qwen Team
- Score
- 66.3%
- Price
- $2.0 input / $6.0 output per 1M tokens
Strengths
- Ranks #1 of 17 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures LongBench v2, not total model capability
- 02ACQwen3.5-397B-A17BAlibaba Cloud / Qwen Team
- Score
- 63.2%
- Price
- $0.60 input / $3.6 output per 1M tokens
Strengths
- Ranks #2 of 17 compared models
- 94th percentile on this benchmark
- C evidence result
Considerations
- This result measures LongBench v2, not total model capability
- 03ACQwen3.6 PlusAlibaba Cloud / Qwen Team
- Score
- 62.0%
- Price
- $0.50 input / $3.0 output per 1M tokens
- Speed
- Up to 16 tok/s via Together
Strengths
- Ranks #3 of 17 compared models
- 88th percentile on this benchmark
- C evidence result
Considerations
- This result measures LongBench v2, not total model capability
- 04NV
- Score
- 61.9%
- Price
- $0.50 input / $2.5 output per 1M tokens
Strengths
- Ranks #4 of 17 compared models
- 81th percentile on this benchmark
- C evidence result
Considerations
- This result measures LongBench v2, not total model capability
- 05MIMiniMax M1 80KMiniMax
- Score
- 61.5%
Strengths
- Ranks #5 of 17 compared models
- 75th percentile on this benchmark
- C evidence result
Considerations
- This result measures LongBench v2, not total model capability
Selection summary
Best AI Models for LongBench v2
Qwen3.8 Max currently leads LongBench v2 with 66.3%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is LongBench v2?
What LongBench v2 measures and how its scores work.
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. This benchmark is not independently verified and has an evidence level of C.
- Family
- LongBench v2
- Modality
- text
- Primary category
- long context
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- longbench-v2|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
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 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.