Benchmarks — published honestly

We ran LongMemEval. We are not #1. Here is why that is the point.

Awareness scores 95.6% recall@5 on LongMemEval_S with hybrid BM25+vector retrieval — running locally on an Apple M1, with zero LLM calls, at 1.7 seconds per query. Two competitors edge us by ~1 point on recall. This page publishes every number, every source, and the scripts to reproduce them.

01 — Our run

Awareness on LongMemEval_S (500 questions)

Environment: Apple M1 / 8GB RAM, all-MiniLM-L6-v2 embedding (23MB ONNX), hybrid RRF (vector 0.7 / BM25 0.3, k=60), session-level indexing. Full run: 843 seconds. Dataset: LongMemEval_S, ~115k tokens and ~47 sessions per question.

Recall@1

77.6%

388 / 500

Recall@3

91.8%

459 / 500

Recall@5

95.6%

478 / 500

Recall@10

97.4%

487 / 500

Ablation — hybrid beats either signal alone

MethodRecall@5
Vector-only92.6%
BM25-only91.4%
Hybrid RRF (0.7 / 0.3)95.6%

By question category (Recall@5)

CategoryR@5Hits
Knowledge update100.0%78/78conflict detection is a first-class feature
Multi-session98.5%131/133cross-session reasoning
Single-session (assistant)98.2%55/56
Temporal reasoning94.7%126/133
Single-session (user)88.6%62/70
Single-session (preference)86.7%26/30

02 — The comparison

Two metrics. We refuse to mix them.

The field quietly conflates two different things: recall@5 (did retrieval surface the right evidence?) and QA accuracy (did the final answer match the judge?). They are not comparable. We publish both, labeled.

Recall@5 — like for like

SystemR@5Source
Memax96.9%docs.memax.app/quickstart/benchmarks
MemPalace (raw ChromaDB)96.6%github.com/MemPalace/mempalace benchmarks
Awareness95.6%this page, scripts public
Stella V5 1.5B (dense)70.0%LongMemEval paper
BM25 (sparse)51.0%LongMemEval paper

We are 1.3 points behind Memax and 1.0 behind MemPalace on recall. We run locally on an M1 with no LLM calls at retrieval; they require a hosted stack. Trade-offs are the honest unit of comparison.

QA accuracy — labeled as a different metric

SystemQAGenerator$/Q
agentmemory96.2%Claude Opus~$0.170/Q
PwC Chronos95.6%enhanced config$0.10–0.54/Q
Mastra OM94.9%GPT-5-mini~$0.130/Q
Backboard93.4%GPT-4.1~$0.070/Q
Ensue93.2%GPT-5-mini~$0.003/Q
OMEGA93.2%GPT-4.1~$0.022/Q
MemMachine93.0%GPT-5-mini~$0.005/Q
Hindsight91.4%Gemini-3-Pro~$0.650/Q
Memax91.2%Claude Sonnet~$0.046/Q
Supermemory85.2%Gemini-3-Pro~$0.110/Q
Zep / Graphiti71.2%GPT-4o~$0.058/Q
Full-context baseline60.2%GPT-4o~$0.290/Q
Mem0 (independent eval)49.0%

Source: aggregated self-reports compiled by docs.mem0-independent comparisons (see footnote). QA accuracy is dominated by the generator model — it measures the answer, not the memory. Awareness does not yet publish a QA number; we will, with the generator and judge pinned.

What the incumbents self-report

LoCoMo

92.5

LongMemEval (QA)

94.4

BEAM 1M

64.1

BEAM 10M

48.6

Mem0's self-reported numbers (mem0.ai blog, April 2026). An independent LongMemEval evaluation of Mem0 measured 49.0% QA accuracy. Self-reported numbers across the field should be read with their source attached — which is what this page does.

03 — Next frontier

LongMemEval-V2 has an official leaderboard. We are submitting.

Released May 2026, LME-V2 raises the bar: 451 curated questions, up to 115M-token histories, and a latency-adjusted frontier score (LAFS) instead of a single number. Published baselines show how much headroom remains:

MethodAccuracy (Small)
No retrieval1.3%
RAG: query → slice42.8%
AgentRunbook-R58.6%
Codex (off-the-shelf)69.9%
AgentRunbook-C74.9%

Our LongMemEval-V2 submission is in preparation. When it lands, the result — whatever it is — will be added to this page with the same disclosure standard.

04 — Full disclosure

Methodology, sources, and limits

Reproduce it yourself

  • Scripts: benchmarks/longmemeval/run_benchmark.py in the open repo
  • Dataset: xiaowu0162/longmemeval-cleaned on HuggingFace
  • Hardware for this run: Apple M1, 8GB RAM — no GPU, no hosted API
  • Embedding: all-MiniLM-L6-v2, 384-d, quantized ONNX

Known limits

  • Single benchmark, single dataset size (LongMemEval_S, not _M).
  • Session-level granularity; turn-level indexing is a known improvement path.
  • Retrieval only — end-to-end QA accuracy not yet published by us.
  • One embedding model tested; stronger models may shift all rows upward.

Why we publish numbers we do not win: memory benchmarks are the only way developers can trust what they are installing. If every vendor only shows the table they top, the field learns nothing. Our advantage is not one point of recall — it is that you can verify everything on this page from an M1 laptop, and that memory itself is free, local-first, and verifiable on-chain. Judge us on that.

Give your AI a memory that actually persists.

Free forever. Local-first. One command: npx @awareness.market/setup

Sources: LongMemEval paper (arxiv 2410.10813) · MemPalace benchmarks (github.com/MemPalace) · Memax benchmark docs (docs.memax.app/quickstart/benchmarks) · Mem0 state-of-memory report (mem0.ai/blog, Apr 2026) · LongMemEval-V2 (arxiv 2605.12493). Competitor figures are self-reported unless labeled independent. Awareness figures are produced by the public scripts in this repository. Last verified: August 2026.

© 2026 Awareness. Memory is free. Intelligence scales.