Unverified50% confidenceBenchmarkExact time
Benchmark tests commonly used to evaluate long-memory systems include RULER, NIAH (Needle-in-a-Haystack), and LongBench.
1
Sources
50%
Confidence
Long-term
Relevance
10/9/2026
First Seen
Sources
Related Entities
Related Claims
Unverified长上下文检索准确率通过Needle-in-a-Haystack测试衡量模型在超长文档中精准定位关键信息的能力76% similarUnverifiedAttSVD's evaluation covers one agent benchmark and the complete LongBench long-context evaluation suite.70% similarUnverifiedThe NIAH test buries key information within a large amount of irrelevant text to examine whether a model can accurately recall it, and is the most intuitive method for measuring long-range information retention.69% similarUnverifiedThe MERIT (Memory Evaluation for Realistic Instrumented Tasks) benchmark measures the actual contribution of memory to task execution and introduces an explicit cost-accounting mechanism for evaluating the long-term memory capabilities of LLM agents.67% similarUnverifiedRULER is a benchmark designed specifically to evaluate long-context model capabilities, covering tasks such as single-hop and multi-hop retrieval, question answering, summarization, and verbatim copying, typically evaluated at context lengths of 16K to 128K.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/992133API
curl https://kongchang.com/api/v1/knowledge/claims/992133MCP
get_claim(id=992133)