Infrastructure & Reliability
DAB v1
Release-gate reliability out of 100%.
Live leaderboard
quaid evals
The benchmark for agent memory systems
Competitive benchmark coverage across reliability, conversational memory, and scale. Quaid’s release gate lives here beside published and estimated peer results.
Latest release
v0.23.0
2026-06-22
DAB v1
99.1%
Release gate score
LoCoMo
20%
Measured today
Published runs
28
Across Quaid evals
Competitive Leaderboard
Grouped bar charts compare Quaid against peer memory systems across the benchmarks that decide whether a memory layer is real or just context stuffing.
Conversational Memory
LoCoMo + LongMemEval
Published and measured dialogue-memory scores.
Scale
BEAM
Extreme-scale memory performance from 100K to 10M tokens.
measured published estimated pending / n/a
Estimated values show an asterisk in tooltips. Pending, not-benchmarked, and n/a states render as ghost bars at 0. GBrain is included from the AI Heroes benchmark, May 2026, which reports an 8.3x win over qmd on 150 real questions. The reported setup uses cloud-based enrichment and is not airgapped; numeric DAB, LoCoMo, LongMemEval, and BEAM scores remain pending until reproducible runs are published.
Methodology
DAB v1
Release-gate scoring and thresholds.
Methodology
LoCoMo
Conversational memory benchmark details.
Methodology
LME
LongMemEval setup and per-type breakdown.
Methodology
BEAM
Extreme-scale memory benchmark methodology.
Methodology
History
Version charts and the last 10 published runs.
Version History
Latest Quaid release-gate runs. Full trend charts live on the history page.
v0.23.0 2026-06-22
99.1%
v0.23.0 2026-06-16
75.8%
v0.22.4 2026-05-18
94.4%
v0.22.3 2026-05-15
94.4%
v0.22.2 2026-05-14
94.4%
Latest MSMARCO snapshot: P@5 17.4% / R@5 38.9%.