agentmemoryleaderboard
A public benchmark that compares agent memory systems using the same procedure
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
It runs smoke tests and formal evaluations against participant-operated Add/Search APIs. Evaluations proceed through ingestion, retrieval, answering, scoring, and aggregation, with progress and error summaries available. It also provides results by track, public rankings, and private evaluation results.
KEY FEATURES
Key features
- Evaluate text, multimodal, and coding agent memory systems by track
- Run synchronous smoke tests on participant-operated Add/Search APIs
- Submit formal evaluation jobs using fixed datasets and Top K
- Show progress and errors for ingestion, retrieval, answering, scoring, and aggregation
- Provide public rankings with filters, dataset columns, and score bars
- View private results tied to an API key and their public review status