long context benchmark
MemoryBench Leaderboard
MemoryBench is an internal recall evaluation over 64K-100K-token conversations containing distractors, conflicting facts, and later updates.
Updated Oct 23, 2026
Models4
Model coverage4
MetricScore
EvidenceB
MemoryBench Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelME | Score85.60% | Percentile100.00% | Participants4 | EvidenceC | Evaluated |
| Rank02 | ModelME | Score85.60% | Percentile66.67% | Participants4 | EvidenceC | Evaluated |
| Rank03 | ModelME | Score85.50% | Percentile33.33% | Participants4 | EvidenceC | Evaluated |
| Rank04 | ModelME | Score85.50% | Percentile0.00% | Participants4 | EvidenceC | Evaluated |
MemoryBench Highlights
The leading models and scores on this benchmark.
MemoryBench Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for MemoryBench
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is MemoryBench?
What MemoryBench measures and how its scores work.
MemoryBench is a long_context benchmark that evaluates recall in conversations ranging from 64K to 100K tokens, using an LLM judge with weighted question-specific rubrics.
It measures Score, reported as a ratio, for recall across conversations with distractors, conflicting facts, and later updates.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- MemoryBench
- Modality
- text
- Primary category
- long context
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about MemoryBench.
Which model scores highest on MemoryBench?
Muse Wonder is currently ranked first with 85.60%.
What are the top three models on MemoryBench?
The current leaders are Muse Wonder (85.60%), Muse Wonder (85.60%), and Muse Wonder Full (85.50%).
Which MemoryBench model has the lowest official input price?
Muse Wonder has the lowest matched official input price at $4.0 input / $20 output per 1M tokens.
Which models are fastest among MemoryBench results?
No matched runtime record is currently available.
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.