{"record": {"aliases": ["agent-memory-benchmark"], "artifacts": [{"id": "gh:datapace-ai/agent-memory-benchmark", "kind": "repo", "url": "https://github.com/datapace-ai/agent-memory-benchmark"}], "categories": ["Agents", "General AI", "Language & Knowledge"], "description": {"en": "Evaluates agent memory systems on a 10-question pilot using LongMemEval, measuring answer accuracy under three judge rules plus token usage, latency, ingestion cost, forgetting curve, and superseded-value handling."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://github.com/datapace-ai/agent-memory-benchmark", "source": "claire_radar", "source_url": "https://github.com/datapace-ai/agent-memory-benchmark", "title": "Agent Memory Benchmark"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:github:datapace-ai/agent-memory-benchmark", "modality": null, "name": "Agent Memory Benchmark", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-candidate", "display_eligible": "true", "origin_record_id": "bm_agent-memory-benchmark_426c2d6c", "origin_source": "github", "recognition_confidence": "0.4", "record_sha256": "d1dba22058803d6d47cc0fd3da5b19bae2eed603299e34279fae585a32f6442d", "relation": "unclear", "review_model": "deepseek-v4-pro", "review_state": "ai-reviewed", "reviewed_at": "2026-09-13T09:41:17.153180Z", "source_url": "https://github.com/datapace-ai/agent-memory-benchmark"}, "publisher": null, "released": "2026-09-08", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:github:datapace-ai/agent-memory-benchmark", "source_url": "https://github.com/datapace-ai/agent-memory-benchmark"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-github-datapace-ai-agent-memory-benchmark", "source": "claire_radar", "source_benchmark_id": "github:datapace-ai/agent-memory-benchmark", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["agent-memory-benchmark"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-09-13", "attemptedAt": "2026-09-13T01:36:51.692703Z", "githubRepo": "https://github.com/datapace-ai/agent-memory-benchmark", "githubScope": "benchmark_repo", "githubStars": 1, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": null, "observedAt": "2026-09-13T01:36:51.692703Z", "signalStatus": {"githubStars": {"observedAt": "2026-09-13T01:36:51.692703Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"state": "not_applicable"}}, "status": "fresh"}, "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "A vendor-neutral, zero-cost benchmark with a defined scoring contract and reproducible Google Colab-style workflow may attract medium attention from developers evaluating memory tools.", "score": 64}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-13T09:41:17.153180Z", "inputHash": "7158c1a2a70204c14d0abcf432e7af34a42143e55308daed2f824f235759a770", "model": "deepseek-v4-pro"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "# Agent Memory Benchmark Built and maintained by [Datapace](https://datapace.ai)", "canonicalNameSource": "official_readme", "decisionReason": "The repository provides a released question set, a documented scoring protocol, and runnable code that allows other teams to reproduce or extend the evaluation.", "evaluationEvidence": {"reuse": "The question set is committed, so no download is needed; `python -m membench.data.download` rebuilds it from LongMemEval.", "scoring": "Accuracy under three published judge rules, tokens per answer, retrieval and answer latency, ingestion cost, a forgetting curve by distance from the evidence, and the rate at which a system serves a superseded value after a fact has changed.", "task": "Every system does the same job: ingest a user's conversation history one session at a time in chronological order, then answer a question using only what it stored."}, "model": "deepseek-v4-pro", "reviewedAt": "2026-09-13T09:41:17.153180Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-candidate", "demo": false, "description": "Evaluates agent memory systems on a 10-question pilot using LongMemEval, measuring answer accuracy under three judge rules plus token usage, latency, ingestion cost, forgetting curve, and superseded-value handling.", "discoverySources": [{"id": "github:datapace-ai/agent-memory-benchmark", "type": "github", "url": "https://github.com/datapace-ai/agent-memory-benchmark"}], "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["discovered via github", "benchmark term in abstract", "no explicit benchmark release evidence"], "snippet": "agent-memory-benchmark Vendor-neutral benchmark of agent memory systems (Mem0, LangMem, Cognee, Graphiti, Letta) against oracle, window and file-search baselines on LongMemEval, at zero API cost # Agent Memory Benchmark A vendor-neutral benchmark for AI agent memory systems, run by Datapace."}, "familyId": "bmf_13dc60f99742", "firstSeenAt": "2026-09-09", "heat": null, "id": "bm_agent-memory-benchmark_426c2d6c", "indexedAt": "2026-09-09T06:18:35.023319Z", "industrySectors": [], "links": {"code": "https://github.com/datapace-ai/agent-memory-benchmark"}, "metrics": [], "motivation": "agent-memory-benchmark Vendor-neutral benchmark of agent memory systems (Mem0, LangMem, Cognee, Graphiti, Letta) against oracle, window and file-search baselines on LongMemEval, at zero API cost # Agent Memory Benchmark A vendor-neutral benchmark for AI agent memory systems, run by Datapace.", "name": "Agent Memory Benchmark", "oneLine": "Evaluates agent memory systems on a 10-question pilot using LongMemEval, measuring answer accuracy under three judge rules plus token usage, latency, ingestion cost, forgetting curve, and superseded-value handling.", "paperTitle": "agent-memory-benchmark", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://github.com/datapace-ai/agent-memory-benchmark", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-09-09T06:18:35.023319Z"}, "publications": [], "ranking": {"30d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 30}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.10323645750017778, "value": 1}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 98, "score": 28}, "method": "attention-ranking-v13", "rank": 98, "score": 28}, "90d": {"confidence": "Low", "coverage": 0.55, "growth": {"components": {"githubStars": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfDatasetDownloads": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}, "hfPaperUpvotes": {"normalization": "percentile", "normalized": null, "percentile": null, "value": null}}, "confidence": "Low", "coverage": 0.0, "elapsedDays": null, "historyDate": null, "rank": null, "score": null, "windowDays": 90}, "level": {"components": {"githubStars": {"normalization": "log-max", "normalized": 0.09340123961313418, "value": 1}, "hfDatasetDownloads": {"normalization": "log-max", "normalized": null, "value": null}, "hfPaperUpvotes": {"normalization": "log-max", "normalized": null, "value": null}}, "confidence": "Low", "coverage": 0.55, "rank": 264, "score": 28}, "method": "attention-ranking-v13", "rank": 264, "score": 28}}, "readiness": "Runnable", "recognitionConfidence": 0.4, "relation": "unclear", "releasedAt": "2026-09-08", "source": {"authors": ["datapace-ai"], "categories": [], "id": "github:datapace-ai/agent-memory-benchmark", "publicSignals": {"githubStars": 1}, "title": "agent-memory-benchmark", "type": "github", "url": "https://github.com/datapace-ai/agent-memory-benchmark"}, "sourceUpdatedAt": "2026-09-09T01:58:02Z", "topics": ["Agents"], "venueAttempts": [], "whyItMatters": "Provides a vendor-neutral, zero-cost methodology for comparing memory frameworks on identical tasks, enabling developers to choose tools based on reproducible local measurements."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
