Mem

Memory systems, readers, and writers.

3 studies
6 comparisons
189 original files

All memory comparisons

Complete + bound counts complete answers with source-matched quotes. Each row is a separate test.

Memory comparison overview
Comparison	Complete + bound	Resources	Evidence scale
Readers · fixed-context replay → Luna / DeepSeek Flash 1017	DeepSeek Flash 101711/13 Luna medium11/13	$0.004811–$0.0193694.54–5.05 s / request	13 inputs / profile2 profiles · one pass
Readers · fixed-context replay → Luna reasoning effort	Luna low10/13 Luna medium11/13 Luna none11/13	$0.004149–$0.0047273.63–4.95 s / request	13 inputs / profile3 profiles · one pass
Systems · native retrieval → Native memory systems	OpenViking11/13 RAGFlow rc111/13	$0.006922–$0.024261Time: withheld from comparison	13 inputs / profile2 profiles · one pass
Systems · nightly version check → Native memory systems	RAGFlow nightly11/13	Cost not reported4.74 s / request	13 inputs / profile1 profile · one pass
Writers · extraction replay → Luna / DeepSeek Flash 1017	Quality not measuredDeepSeek Flash 1017 · Luna medium	$0.001142–$0.0029144.57–8.31 s / request	3 inputs / profile2 profiles · one pass
Writers · extraction replay → Luna reasoning effort Viewing	Quality not measuredLuna low · Luna medium · Luna none	$0.000709–$0.0011203.52–7.72 s / request	4 inputs / profile3 profiles · one pass

Separate tests · no combined ranking.

说明

Costs are workload totals in USD; resource ranges span profiles within that row. Quote binding does not establish semantic entailment.

Writers · extraction replay

Luna reasoning effort · Extraction prompt replay

↑ All comparisons

Refine comparison

Study

Test

4 inputs / profile · 3 profiles · October 20, 2026

Test method

Resource use for the same inputs. Semantic writer quality was not measured.

Results & evidence ↓

Mean time

Seconds / input · lower is better

Luna low6.11 s
Luna medium7.72 s
Luna none3.52 s

Settled cost

USD / test · lower is better

Luna low$0.000999
Luna medium$0.001120
Luna none$0.000709

Results and evidence

Sort columns. Open Details for the profile and original record.

Expand a row for all metrics

Sort

Metric

Writer · 4 inputs. Values are specific to this test. Actual settled USD for four extraction requests. Atom count and JSON parsing are descriptive, not semantic quality scores.
Profile version						Evidence
Luna nonegpt-6-luna	4	3.518 s	$0.000709	13	4/4	Result evidence Luna none Writer · 4 inputs · 4 inputs per profile Profile Time s · mean Cost USD total Luna low 6.107 s $0.000999 Luna medium 7.720 s $0.001120 Luna noneSelected 3.518 s $0.000709 Resource use for saved extraction outputs. Writer quality is not scored. All recorded measures Inputs per profile 4 Time s · mean 3.518 s Cost USD total $0.000709100 Atoms count 13 JSON outputs 4/4 Configuration & original record Recorded date 2026-10-20 Model openai/gpt-6-luna Provider OpenAI; no fallback Effort none OutputAllowance 128,000 tokens Extraction prompt replay Actual settled USD for four extraction requests. Atom count and JSON parsing are descriptive, not semantic quality scores. One diagnostic pass with real provider calls. Sort order is descriptive. Source binding alone does not establish semantic support. Open original metrics.json ↗
Luna lowgpt-6-luna	4	6.107 s	$0.000999	11	4/4	Result evidence Luna low Writer · 4 inputs · 4 inputs per profile Profile Time s · mean Cost USD total Luna lowSelected 6.107 s $0.000999 Luna medium 7.720 s $0.001120 Luna none 3.518 s $0.000709 Resource use for saved extraction outputs. Writer quality is not scored. All recorded measures Inputs per profile 4 Time s · mean 6.107 s Cost USD total $0.000998600 Atoms count 11 JSON outputs 4/4 Configuration & original record Recorded date 2026-10-20 Model openai/gpt-6-luna Provider OpenAI; no fallback Effort low OutputAllowance 128,000 tokens Extraction prompt replay Actual settled USD for four extraction requests. Atom count and JSON parsing are descriptive, not semantic quality scores. One diagnostic pass with real provider calls. Sort order is descriptive. Source binding alone does not establish semantic support. Open original metrics.json ↗
Luna mediumgpt-6-luna	4	7.720 s	$0.001120	9	4/4	Result evidence Luna medium Writer · 4 inputs · 4 inputs per profile Profile Time s · mean Cost USD total Luna low 6.107 s $0.000999 Luna mediumSelected 7.720 s $0.001120 Luna none 3.518 s $0.000709 Resource use for saved extraction outputs. Writer quality is not scored. All recorded measures Inputs per profile 4 Time s · mean 7.720 s Cost USD total $0.001119600 Atoms count 9 JSON outputs 4/4 Configuration & original record Recorded date 2026-10-20 Model openai/gpt-6-luna Provider OpenAI; no fallback Effort medium OutputAllowance 128,000 tokens Extraction prompt replay Actual settled USD for four extraction requests. Atom count and JSON parsing are descriptive, not semantic quality scores. One diagnostic pass with real provider calls. Sort order is descriptive. Source binding alone does not establish semantic support. Open original metrics.json ↗

One pass · — Not measured.

说明

USD totals are rounded to 6 decimals. The value tooltip and original record retain the recorded precision. Sorting is descriptive.

Actual settled USD for four extraction requests. Atom count and JSON parsing are descriptive, not semantic quality scores.

Recorded finding

Four writer inputs per profile; all outputs parsed as direct JSON. Returned atom counts do not measure semantic quality.

Protocol and metric definitions

Four identical extraction inputs, including English and Chinese conditional decisions. All profiles preserve the two tested conditional-scope and owner samples. Subject and effective-time omissions remain.

Synthetic inputs · real provider calls · one pass · no repeated-trial interval.

Inputs: paired extraction requests. Time: mean request duration. Cost: settled USD for the selected workload. Output counts and JSON parsing are not quality scores.

Full conditions and limitations

Results