iiRecord
Agentic AI Atlas · MemoryBench
page:docs-supermemory-research-raw-12-memorybencha5c.ai
II.
Page reference

page:docs-supermemory-research-raw-12-memorybench

Reading · 1 min

MemoryBench reference

Source: https://supermemory.ai/docs/memorybench/overview

Pagewiki/docs/supermemory-research/raw/12-memorybench.mdOutgoing · 0Incoming · 1

MemoryBench

Source: https://supermemory.ai/docs/memorybench/overview

Core Purpose

MemoryBench is an open-source framework for standardized, reproducible benchmarks of memory layer providers. Designed for rigorous, accessible, and industry-aligned evaluations.

Key Features

  • Design custom evaluations
  • Run industry-standard benchmarks across memory providers
  • Transparent and reproducible results

Components

  • Installation guide
  • CLI Reference
  • Architecture documentation
  • Web Interface: bun run src/index.ts serve
  • MemScore metric for comparing quality, latency, and token efficiency

Supported Providers

  • Supermemory (cloud-based memory layer)
  • Mem0 (graph-based memory)
  • Zep (long-term memory for AI)

Additional providers being integrated with community contribution opportunities.