LongBench v2

LongBench v2 is a benchmark for reading and reasoning over very long inputs. The official description says: "LongBench v2 - long context understanding benchmark". Below we describe in simple words what benchmarks of this type generally examine.

LongBench v2 measures reading and reasoning over very long inputs. In detail, it checks:

• finding relevant facts buried in documents hundreds of thousands of tokens long
• retrieving multiple needles hidden across the full context
• reasoning over whole books, codebases or conversation histories
• staying accurate near the context window limit (no 'lost in the middle' falloff)

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.

Official methodology note: LongBench v2 - long context understanding benchmark.

What is the LongBench v2 benchmark?

LongBench v2 is a benchmark for reading and reasoning over very long inputs. The official description says: 'LongBench v2 - long context understanding benchmark'. Below we describe in simple words what benchmarks of this type generally examine.

What does LongBench v2 test?

LongBench v2 mainly measures reading and reasoning over very long inputs. Concretely, it checks: finding relevant facts buried in documents hundreds of thousands of tokens long; retrieving multiple needles hidden across the full context; reasoning over whole books, codebases or conversation histories.

How is the LongBench v2 score calculated?

The score is accuracy: the percentage of tasks answered or solved correctly. 100% means every task was solved perfectly; higher is better.

Where do I get LongBench v2 data and scores?

The official benchmark page/reference is linked on this page. Every score on Open LLM Radar is also available as open data: download the scores JSON (http://open-llm-radar-api:8000/api/v1/benchmarks/longbench-v2/scores) or pull it directly from the Open LLM Radar API with any HTTP client.