Benchmark RadarAI BENCHMARK PROFILE
StreamMemBench
StreamMemBench is a streaming benchmark for evaluating agent memory with two-step task sequences around evidence anchors from EgoLife egocentric streams, with four diagnostic metrics.
- Released
- 2026-06-12
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Existing memory benchmarks test recall or task improvement in isolation. StreamMemBench evaluates the trajectory from streaming observations to future-oriented assistance, diagnosing whether memory systems use evidence and incorporate feedback.
Motivation
A central role of personal-agent memory is to turn stored information and prior interactions into future-oriented assistance.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.