{
  "claims": [
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Mem0",
      "scope": "April 2026 algorithm on the managed platform, which includes proprietary optimizations; single pass, top_200; was 71.4. Judge not stated.",
      "source": {
        "commit": "c93420c49a6b14c3d446bdb156d96811908fd90a",
        "line": 49,
        "retrieved": "2026-10-06",
        "title": "Mem0 README",
        "url": "https://github.com/mem0ai/mem0/blob/c93420c49a6b14c3d446bdb156d96811908fd90a/README.md#L49"
      },
      "split": "",
      "system": "Mem0",
      "unit": "%",
      "value": 92.5,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Mem0",
      "scope": "Same setup; was 67.8. Judge not stated.",
      "source": {
        "commit": "c93420c49a6b14c3d446bdb156d96811908fd90a",
        "line": 50,
        "retrieved": "2026-10-06",
        "title": "Mem0 README",
        "url": "https://github.com/mem0ai/mem0/blob/c93420c49a6b14c3d446bdb156d96811908fd90a/README.md#L50"
      },
      "split": "",
      "system": "Mem0",
      "unit": "%",
      "value": 94.4,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Mem0",
      "scope": "Same setup. Judge not stated.",
      "source": {
        "commit": "c93420c49a6b14c3d446bdb156d96811908fd90a",
        "line": 51,
        "retrieved": "2026-10-06",
        "title": "Mem0 README",
        "url": "https://github.com/mem0ai/mem0/blob/c93420c49a6b14c3d446bdb156d96811908fd90a/README.md#L51"
      },
      "split": "1M",
      "system": "Mem0",
      "unit": "%",
      "value": 64.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Mem0",
      "scope": "Same setup. Judge not stated.",
      "source": {
        "commit": "c93420c49a6b14c3d446bdb156d96811908fd90a",
        "line": 52,
        "retrieved": "2026-10-06",
        "title": "Mem0 README",
        "url": "https://github.com/mem0ai/mem0/blob/c93420c49a6b14c3d446bdb156d96811908fd90a/README.md#L52"
      },
      "split": "10M",
      "system": "Mem0",
      "unit": "%",
      "value": 48.6,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "Mem0",
      "scope": "Mem0 paper.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0",
      "unit": "%",
      "value": 66.88,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "Mem0",
      "scope": "Mem0 paper.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0 graph",
      "unit": "%",
      "value": 68.44,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "Zep",
      "scope": "Mem0 authors running Zep. Zep disputes this number (see its blog row).",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Zep",
      "unit": "%",
      "value": 65.99,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "LangMem",
      "scope": "Mem0 paper.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "LangMem",
      "unit": "%",
      "value": 58.1,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "OpenAI memory",
      "scope": "OpenAI memory as run by the Mem0 authors; closed source.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "OpenAI memory",
      "unit": "%",
      "value": 52.9,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "A-Mem",
      "scope": "Mem0 paper.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "A-Mem",
      "unit": "%",
      "value": 48.38,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "Full context (no memory)",
      "scope": "Whole conversation in the prompt, about 26,000 tokens.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Full context (no memory)",
      "unit": "%",
      "value": 72.9,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "Mem0",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0",
      "unit": "s",
      "value": 0.708,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "Mem0",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0",
      "unit": "s",
      "value": 1.44,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "Mem0",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0 graph",
      "unit": "s",
      "value": 1.091,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "Mem0",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Mem0 graph",
      "unit": "s",
      "value": 2.59,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "Zep",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Zep",
      "unit": "s",
      "value": 1.292,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "Zep",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Zep",
      "unit": "s",
      "value": 2.926,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "LangMem",
      "scope": "Search alone was 17.99 s; the total is search plus answer.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "LangMem",
      "unit": "s",
      "value": 18.53,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "LangMem",
      "scope": "Search alone was 59.82 s.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "LangMem",
      "unit": "s",
      "value": 60.4,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "Full context (no memory)",
      "scope": "Whole conversation in the prompt.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Full context (no memory)",
      "unit": "s",
      "value": 9.87,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "Full context (no memory)",
      "scope": "Whole conversation in the prompt.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "Full context (no memory)",
      "unit": "s",
      "value": 17.117,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "A-Mem",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "A-Mem",
      "unit": "s",
      "value": 1.41,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "A-Mem",
      "scope": "Search plus answer, seconds.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "A-Mem",
      "unit": "s",
      "value": 4.374,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p50",
      "product": "OpenAI memory",
      "scope": "No search step; memories are extracted manually in the prompt.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "OpenAI memory",
      "unit": "s",
      "value": 0.466,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "latency",
      "group": "mem0-paper",
      "high": null,
      "low": null,
      "metric": "total latency p95",
      "product": "OpenAI memory",
      "scope": "No search step.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mem0 paper, arXiv 2504.19413",
        "url": "https://arxiv.org/abs/2504.19413"
      },
      "split": "",
      "system": "OpenAI memory",
      "unit": "s",
      "value": 0.889,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "J score (LLM judge)",
      "product": "Zep",
      "scope": "Zep corrected its own earlier post and reports 75.14 +/- 0.17, against 65.99 in the Mem0 paper.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Zep blog, corrected LoCoMo result",
        "url": "https://blog.getzep.com/lies-damn-lies-statistics-is-mem0-really-sota-in-agent-memory/"
      },
      "split": "",
      "system": "Zep",
      "unit": "%",
      "value": 75.14,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "zep-paper",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Zep",
      "scope": "Zep paper; reader gpt-4o-mini; full-context baseline 55.4; latency 3.20 s against 31.3 s.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Zep paper, arXiv 2501.13956, Table 2",
        "url": "https://arxiv.org/abs/2501.13956"
      },
      "split": "S",
      "system": "Zep",
      "unit": "%",
      "value": 63.8,
      "variant": "gpt-4o-mini reader",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "zep-paper",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Zep",
      "scope": "Zep paper; reader gpt-4o; full-context baseline 60.2; latency 2.58 s against 28.9 s.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Zep paper, arXiv 2501.13956, Table 2",
        "url": "https://arxiv.org/abs/2501.13956"
      },
      "split": "S",
      "system": "Zep",
      "unit": "%",
      "value": 71.2,
      "variant": "gpt-4o reader",
      "who": "self"
    },
    {
      "baseline": true,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "zep-paper",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Full context (no memory)",
      "scope": "Zep paper; reader gpt-4o-mini; about 115k tokens of context.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Zep paper, arXiv 2501.13956, Table 2",
        "url": "https://arxiv.org/abs/2501.13956"
      },
      "split": "S",
      "system": "Full context (no memory)",
      "unit": "%",
      "value": 55.4,
      "variant": "gpt-4o-mini reader",
      "who": "self"
    },
    {
      "baseline": true,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "zep-paper",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Full context (no memory)",
      "scope": "Zep paper; reader gpt-4o; about 115k tokens of context.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Zep paper, arXiv 2501.13956, Table 2",
        "url": "https://arxiv.org/abs/2501.13956"
      },
      "split": "S",
      "system": "Full context (no memory)",
      "unit": "%",
      "value": 60.2,
      "variant": "gpt-4o reader",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@5",
      "product": "MemPalace",
      "scope": "500 questions; raw semantic search, no LLM, no heuristics. Recall variant (any versus all gold sessions) not stated in the README table.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 245,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L245"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 96.6,
      "variant": "raw",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@5",
      "product": "MemPalace",
      "scope": "Hybrid v4, held-out 450 questions; tuned on 50 dev questions.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 246,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L246"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 98.4,
      "variant": "hybrid v4, held out",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@10",
      "product": "MemPalace",
      "scope": "Session level, top-10, no rerank; 1,986 questions.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 266,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L266"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 60.3,
      "variant": "session",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@10",
      "product": "MemPalace",
      "scope": "Hybrid v5, top-10, no rerank; same 1,986 questions.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 267,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L267"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 88.9,
      "variant": "hybrid v5",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "ConvoMem",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "average recall",
      "product": "MemPalace",
      "scope": "All categories, 250 items, 50 per category.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 268,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L268"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 92.9,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "MemBench",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@5",
      "product": "MemPalace",
      "scope": "ACL 2025, 8,500 items, all categories.",
      "source": {
        "commit": "35dc62153c693af199e69acb91942a8afc0ebcc7",
        "line": 269,
        "retrieved": "2026-10-06",
        "title": "MemPalace README",
        "url": "https://github.com/MemPalace/mempalace/blob/35dc62153c693af199e69acb91942a8afc0ebcc7/README.md#L269"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 80.3,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@5",
      "product": "agentmemory",
      "scope": "500 questions; all-MiniLM-L6-v2 embeddings, local. R@10 98.6, MRR 88.2.",
      "source": {
        "commit": "007a1a7fe8646a03d8652eb0712400cc6f0fcca3",
        "line": 356,
        "retrieved": "2026-10-06",
        "title": "agentmemory README",
        "url": "https://github.com/rohitg00/agentmemory/blob/007a1a7fe8646a03d8652eb0712400cc6f0fcca3/README.md#L356"
      },
      "split": "S",
      "system": "agentmemory",
      "unit": "%",
      "value": 95.2,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": true,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "R@5",
      "product": "agentmemory",
      "scope": "agentmemory's own keyword-only baseline on the same questions.",
      "source": {
        "commit": "007a1a7fe8646a03d8652eb0712400cc6f0fcca3",
        "line": 357,
        "retrieved": "2026-10-06",
        "title": "agentmemory README",
        "url": "https://github.com/rohitg00/agentmemory/blob/007a1a7fe8646a03d8652eb0712400cc6f0fcca3/README.md#L357"
      },
      "split": "S",
      "system": "BM25-only fallback",
      "unit": "%",
      "value": 86.2,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@15",
      "product": "supermemory",
      "scope": "Adds about 720 tokens of context, a 99.4% reduction. At a different k from the other recall rows.",
      "source": {
        "commit": "3535ff700da85134d1867e5eac1b649e8735e3af",
        "line": 357,
        "retrieved": "2026-10-06",
        "title": "supermemory README",
        "url": "https://github.com/supermemoryai/supermemory/blob/3535ff700da85134d1867e5eac1b649e8735e3af/README.md#L357"
      },
      "split": "",
      "system": "supermemory",
      "unit": "%",
      "value": 95.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "unspecified",
      "group": "",
      "high": null,
      "low": null,
      "metric": "ranking claim: #1",
      "product": "supermemory",
      "scope": "States #1 with no figure in the README.",
      "source": {
        "commit": "3535ff700da85134d1867e5eac1b649e8735e3af",
        "line": 353,
        "retrieved": "2026-10-06",
        "title": "supermemory README",
        "url": "https://github.com/supermemoryai/supermemory/blob/3535ff700da85134d1867e5eac1b649e8735e3af/README.md#L353"
      },
      "split": "",
      "system": "supermemory",
      "unit": "%",
      "value": null,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "unspecified",
      "group": "",
      "high": null,
      "low": null,
      "metric": "ranking claim: #1",
      "product": "supermemory",
      "scope": "States #1 with no figure in the README.",
      "source": {
        "commit": "3535ff700da85134d1867e5eac1b649e8735e3af",
        "line": 354,
        "retrieved": "2026-10-06",
        "title": "supermemory README",
        "url": "https://github.com/supermemoryai/supermemory/blob/3535ff700da85134d1867e5eac1b649e8735e3af/README.md#L354"
      },
      "split": "",
      "system": "supermemory",
      "unit": "%",
      "value": null,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "ConvoMem",
      "family": "unspecified",
      "group": "",
      "high": null,
      "low": null,
      "metric": "ranking claim: #1",
      "product": "supermemory",
      "scope": "States #1 with no figure in the README.",
      "source": {
        "commit": "3535ff700da85134d1867e5eac1b649e8735e3af",
        "line": 355,
        "retrieved": "2026-10-06",
        "title": "supermemory README",
        "url": "https://github.com/supermemoryai/supermemory/blob/3535ff700da85134d1867e5eac1b649e8735e3af/README.md#L355"
      },
      "split": "",
      "system": "supermemory",
      "unit": "%",
      "value": null,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "strict recall_all@5",
      "product": "gbrain",
      "scope": "451 of 470 questions; every required session must be in the top five; opaque session ids; with Voyage reranker.",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 49,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L49"
      },
      "split": "",
      "system": "gbrain",
      "unit": "%",
      "value": 95.96,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "strict recall_all@5",
      "product": "gbrain",
      "scope": "434 of 470; same without the reranker.",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 88,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L88"
      },
      "split": "",
      "system": "gbrain",
      "unit": "%",
      "value": 92.34,
      "variant": "no reranker",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "gbrain-recount",
      "high": null,
      "low": null,
      "metric": "strict recall_all@5",
      "product": "MemPalace",
      "scope": "gbrain authors recounting MemPalace's saved rankings strictly (423 of 470); hybrid search plus LLM rerank.",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 89,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L89"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 90.0,
      "variant": "hybrid + LLM rerank, recounted",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "gbrain-recount",
      "high": null,
      "low": null,
      "metric": "strict recall_all@5",
      "product": "MemPalace",
      "scope": "Recount of the raw vector search (403 of 470).",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 91,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L91"
      },
      "split": "",
      "system": "MemPalace",
      "unit": "%",
      "value": 85.7,
      "variant": "raw, recounted",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "gbrain",
      "scope": "453 of 500; house reader with reranker.",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 50,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L50"
      },
      "split": "",
      "system": "gbrain",
      "unit": "%",
      "value": 90.6,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "gbrain",
      "scope": "447 of 500; gpt-5.4 reader on gbrain retrieval.",
      "source": {
        "commit": "48dd47bd8206e5f1caa6e6f23746d209e9b88b2e",
        "line": 51,
        "retrieved": "2026-10-06",
        "title": "gbrain-evals README",
        "url": "https://github.com/garrytan/gbrain-evals/blob/48dd47bd8206e5f1caa6e6f23746d209e9b88b2e/README.md#L51"
      },
      "split": "",
      "system": "gbrain",
      "unit": "%",
      "value": 89.4,
      "variant": "gpt-5.4 reader",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "overall accuracy",
      "product": "Memori",
      "scope": "About 721 tokens per query, 2.8% of the full-context footprint.",
      "source": {
        "commit": "574b1ea3e876f100ef82c37817d603eb7e258e59",
        "line": 130,
        "retrieved": "2026-10-06",
        "title": "Memori README",
        "url": "https://github.com/MemoriLabs/Memori/blob/574b1ea3e876f100ef82c37817d603eb7e258e59/README.md#L130"
      },
      "split": "",
      "system": "Memori",
      "unit": "%",
      "value": 87.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "average accuracy",
      "product": "CORE",
      "scope": "Across single-hop, multi-hop, open-domain and temporal questions.",
      "source": {
        "commit": "4a5b18d8db55d66e5dfda41b18461f359b340c42",
        "line": 244,
        "retrieved": "2026-10-06",
        "title": "CORE README",
        "url": "https://github.com/RedPlanetHQ/core/blob/4a5b18d8db55d66e5dfda41b18461f359b340c42/README.md#L244"
      },
      "split": "",
      "system": "CORE",
      "unit": "%",
      "value": 88.24,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "overall accuracy",
      "product": "ByteRover",
      "scope": "1,982 questions, 272 documents; production codebase, no separate prototype.",
      "source": {
        "commit": "1052ac1a5dd0fde4da8693d4712064f7876c269c",
        "line": 55,
        "retrieved": "2026-10-06",
        "title": "ByteRover README",
        "url": "https://github.com/campfirein/byterover-cli/blob/1052ac1a5dd0fde4da8693d4712064f7876c269c/README.md#L55"
      },
      "split": "",
      "system": "ByteRover",
      "unit": "%",
      "value": 96.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "overall accuracy",
      "product": "ByteRover",
      "scope": "500 questions, 23,867 documents.",
      "source": {
        "commit": "1052ac1a5dd0fde4da8693d4712064f7876c269c",
        "line": 57,
        "retrieved": "2026-10-06",
        "title": "ByteRover README",
        "url": "https://github.com/campfirein/byterover-cli/blob/1052ac1a5dd0fde4da8693d4712064f7876c269c/README.md#L57"
      },
      "split": "S",
      "system": "ByteRover",
      "unit": "%",
      "value": 92.8,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "MemOS",
      "scope": "Evaluated through OmniMemEval, a framework run by the same organisation.",
      "source": {
        "commit": "a7367d07e55db61099f7b4e2c1108bc5831a24f3",
        "line": 78,
        "retrieved": "2026-10-06",
        "title": "MemOS README",
        "url": "https://github.com/MemTensor/MemOS/blob/a7367d07e55db61099f7b4e2c1108bc5831a24f3/README.md#L78"
      },
      "split": "",
      "system": "MemOS",
      "unit": "%",
      "value": 88.83,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "MemOS",
      "scope": "Evaluated through OmniMemEval.",
      "source": {
        "commit": "a7367d07e55db61099f7b4e2c1108bc5831a24f3",
        "line": 79,
        "retrieved": "2026-10-06",
        "title": "MemOS README",
        "url": "https://github.com/MemTensor/MemOS/blob/a7367d07e55db61099f7b4e2c1108bc5831a24f3/README.md#L79"
      },
      "split": "",
      "system": "MemOS",
      "unit": "%",
      "value": 89.2,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "PersonaMem v2",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "MemOS",
      "scope": "Evaluated through OmniMemEval.",
      "source": {
        "commit": "a7367d07e55db61099f7b4e2c1108bc5831a24f3",
        "line": 80,
        "retrieved": "2026-10-06",
        "title": "MemOS README",
        "url": "https://github.com/MemTensor/MemOS/blob/a7367d07e55db61099f7b4e2c1108bc5831a24f3/README.md#L80"
      },
      "split": "",
      "system": "MemOS",
      "unit": "%",
      "value": 40.58,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "HaluMem",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "MemOS",
      "scope": "Evaluated through OmniMemEval.",
      "source": {
        "commit": "a7367d07e55db61099f7b4e2c1108bc5831a24f3",
        "line": 81,
        "retrieved": "2026-10-06",
        "title": "MemOS README",
        "url": "https://github.com/MemTensor/MemOS/blob/a7367d07e55db61099f7b4e2c1108bc5831a24f3/README.md#L81"
      },
      "split": "",
      "system": "MemOS",
      "unit": "%",
      "value": 80.91,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "MemOS",
      "scope": "Evaluated through OmniMemEval.",
      "source": {
        "commit": "a7367d07e55db61099f7b4e2c1108bc5831a24f3",
        "line": 82,
        "retrieved": "2026-10-06",
        "title": "MemOS README",
        "url": "https://github.com/MemTensor/MemOS/blob/a7367d07e55db61099f7b4e2c1108bc5831a24f3/README.md#L82"
      },
      "split": "10M",
      "system": "MemOS",
      "unit": "%",
      "value": 56.75,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "agentic score",
      "product": "ReMe",
      "scope": "500 questions.",
      "source": {
        "commit": "084c02e43a885d5d42a36c39ce9bd78263c23247",
        "line": 355,
        "retrieved": "2026-10-06",
        "title": "ReMe README",
        "url": "https://github.com/agentscope-ai/ReMe/blob/084c02e43a885d5d42a36c39ce9bd78263c23247/README.md#L355"
      },
      "split": "cleaned-s",
      "system": "ReMe",
      "unit": "%",
      "value": 89.4,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "agentic score",
      "product": "ReMe",
      "scope": "20 cases, 400 questions.",
      "source": {
        "commit": "084c02e43a885d5d42a36c39ce9bd78263c23247",
        "line": 356,
        "retrieved": "2026-10-06",
        "title": "ReMe README",
        "url": "https://github.com/agentscope-ai/ReMe/blob/084c02e43a885d5d42a36c39ce9bd78263c23247/README.md#L356"
      },
      "split": "100K",
      "system": "ReMe",
      "unit": "%",
      "value": 66.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "agentic score",
      "product": "ReMe",
      "scope": "35 cases, 700 questions.",
      "source": {
        "commit": "084c02e43a885d5d42a36c39ce9bd78263c23247",
        "line": 357,
        "retrieved": "2026-10-06",
        "title": "ReMe README",
        "url": "https://github.com/agentscope-ai/ReMe/blob/084c02e43a885d5d42a36c39ce9bd78263c23247/README.md#L357"
      },
      "split": "1M",
      "system": "ReMe",
      "unit": "%",
      "value": 65.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "unspecified",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Memanto",
      "scope": "README calls these public recall benchmarks but does not state the metric, and warns cross-project scores are not comparable.",
      "source": {
        "commit": "aac858c0d918e8194b8626a8abe5ef0af94de7a9",
        "line": 332,
        "retrieved": "2026-10-06",
        "title": "Memanto README",
        "url": "https://github.com/moorcheh-ai/memanto/blob/aac858c0d918e8194b8626a8abe5ef0af94de7a9/README.md#L332"
      },
      "split": "",
      "system": "Memanto",
      "unit": "%",
      "value": 89.8,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "unspecified",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score",
      "product": "Memanto",
      "scope": "Same caveat.",
      "source": {
        "commit": "aac858c0d918e8194b8626a8abe5ef0af94de7a9",
        "line": 332,
        "retrieved": "2026-10-06",
        "title": "Memanto README",
        "url": "https://github.com/moorcheh-ai/memanto/blob/aac858c0d918e8194b8626a8abe5ef0af94de7a9/README.md#L332"
      },
      "split": "",
      "system": "Memanto",
      "unit": "%",
      "value": 87.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score (0 to 1 scale)",
      "product": "Cognee",
      "scope": "Reported 0.79: four rounds over 20 questions from one held-out conversation. Shown here as 79.",
      "source": {
        "commit": "b32d8afc59e1064d9291b9828a8a147be9cc8bab",
        "line": 291,
        "retrieved": "2026-10-06",
        "title": "Cognee README",
        "url": "https://github.com/topoteretes/cognee/blob/b32d8afc59e1064d9291b9828a8a147be9cc8bab/README.md#L291"
      },
      "split": "100K",
      "system": "Cognee",
      "unit": "%",
      "value": 79.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "score (0 to 1 scale)",
      "product": "Cognee",
      "scope": "Reported 0.67: exploratory, question-type routing selected and scored on the same questions. Shown here as 67.",
      "source": {
        "commit": "b32d8afc59e1064d9291b9828a8a147be9cc8bab",
        "line": 292,
        "retrieved": "2026-10-06",
        "title": "Cognee README",
        "url": "https://github.com/topoteretes/cognee/blob/b32d8afc59e1064d9291b9828a8a147be9cc8bab/README.md#L292"
      },
      "split": "10M",
      "system": "Cognee",
      "unit": "%",
      "value": 67.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "LLM score",
      "product": "Nemori",
      "scope": "Reported 0.8305 overall, version V5; 1,540 questions across the four categories.",
      "source": {
        "commit": "d2a6dff6e5481214a0be6a2d10147feccfc16244",
        "line": 171,
        "retrieved": "2026-10-06",
        "title": "Nemori README",
        "url": "https://github.com/nemori-ai/nemori/blob/d2a6dff6e5481214a0be6a2d10147feccfc16244/README.md#L171"
      },
      "split": "",
      "system": "Nemori",
      "unit": "%",
      "value": 83.05,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "",
      "high": 83.0,
      "low": 80.0,
      "metric": "accuracy",
      "product": "OpenViking",
      "scope": "Reports 80 to 83% across three agent integrations, against 24 to 57% on their native memory; reader Doubao 2.0 Pro.",
      "source": {
        "commit": "10f368145f43e672486e3c692981ee8cc97a581a",
        "line": 120,
        "retrieved": "2026-10-06",
        "title": "OpenViking README",
        "url": "https://github.com/volcengine/OpenViking/blob/10f368145f43e672486e3c692981ee8cc97a581a/README.md#L120"
      },
      "split": "",
      "system": "OpenViking",
      "unit": "%",
      "value": null,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "FullText",
      "scope": "LightMem authors; backbone and judge gpt-4o-mini.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 393,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L393"
      },
      "split": "",
      "system": "FullText",
      "unit": "%",
      "value": 73.83,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "NaiveRAG",
      "scope": "LightMem authors; backbone and judge gpt-4o-mini.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 394,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L394"
      },
      "split": "",
      "system": "NaiveRAG",
      "unit": "%",
      "value": 63.64,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "A-Mem",
      "scope": "LightMem authors; backbone and judge gpt-4o-mini.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 395,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L395"
      },
      "split": "",
      "system": "A-Mem",
      "unit": "%",
      "value": 64.16,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "MemoryOS",
      "scope": "LightMem authors; the evaluation build.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 396,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L396"
      },
      "split": "",
      "system": "MemoryOS",
      "unit": "%",
      "value": 58.25,
      "variant": "eval build",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "MemoryOS",
      "scope": "LightMem authors; the PyPI release.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 397,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L397"
      },
      "split": "",
      "system": "MemoryOS",
      "unit": "%",
      "value": 54.87,
      "variant": "PyPI",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "Mem0",
      "scope": "LightMem authors; the open-source package.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 398,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L398"
      },
      "split": "",
      "system": "Mem0",
      "unit": "%",
      "value": 36.49,
      "variant": "open source",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "Mem0",
      "scope": "LightMem authors; the hosted API.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 399,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L399"
      },
      "split": "",
      "system": "Mem0",
      "unit": "%",
      "value": 61.69,
      "variant": "API",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "lightmem",
      "high": null,
      "low": null,
      "metric": "ACC",
      "product": "Mem0",
      "scope": "LightMem authors; the hosted API with graph.",
      "source": {
        "commit": "8449d574df6bae1bdf3314a1564da65e2f37e046",
        "line": 400,
        "retrieved": "2026-10-06",
        "title": "LightMem README",
        "url": "https://github.com/zjunlp/LightMem/blob/8449d574df6bae1bdf3314a1564da65e2f37e046/README.md#L400"
      },
      "split": "",
      "system": "Mem0 graph",
      "unit": "%",
      "value": 60.32,
      "variant": "API",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark, operated by Vectorize, which makes Hindsight.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "S",
      "system": "Hindsight",
      "unit": "%",
      "value": 94.6,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark; 1,986 questions.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "",
      "system": "Hindsight",
      "unit": "%",
      "value": 92.0,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "PersonaMem",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "32k",
      "system": "Hindsight",
      "unit": "%",
      "value": 86.6,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LifeBench",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "en",
      "system": "Hindsight",
      "unit": "%",
      "value": 71.5,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark comparison table. Its own results page shows 75.0 for single-query mode.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "100K",
      "system": "Hindsight",
      "unit": "%",
      "value": 86.2,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark comparison table. Its own results page shows 73.9 for single-query mode.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "1M",
      "system": "Hindsight",
      "unit": "%",
      "value": 79.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Agent Memory Benchmark and the Hindsight results page agree.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "10M",
      "system": "Hindsight",
      "unit": "%",
      "value": 64.1,
      "variant": "",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Hindsight results page, single-query mode.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Hindsight benchmarks page",
        "url": "https://benchmarks.hindsight.vectorize.io/"
      },
      "split": "100K",
      "system": "Hindsight",
      "unit": "%",
      "value": 75.0,
      "variant": "single query",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "BEAM",
      "family": "answer",
      "group": "",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Hindsight",
      "scope": "Hindsight results page, single-query mode.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Hindsight benchmarks page",
        "url": "https://benchmarks.hindsight.vectorize.io/"
      },
      "split": "1M",
      "system": "Hindsight",
      "unit": "%",
      "value": 73.9,
      "variant": "single query",
      "who": "self"
    },
    {
      "baseline": false,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Cognee",
      "scope": "Agent Memory Benchmark, run by Hindsight's maker.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "",
      "system": "Cognee",
      "unit": "%",
      "value": 80.3,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "PersonaMem",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "Cognee",
      "scope": "Agent Memory Benchmark, run by Hindsight's maker.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "32k",
      "system": "Cognee",
      "unit": "%",
      "value": 81.8,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LoCoMo",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "hybrid-search",
      "scope": "Agent Memory Benchmark baseline: plain hybrid search.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "",
      "system": "hybrid-search",
      "unit": "%",
      "value": 79.1,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LongMemEval",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "hybrid-search",
      "scope": "Agent Memory Benchmark baseline: plain hybrid search.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "S",
      "system": "hybrid-search",
      "unit": "%",
      "value": 74.0,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "PersonaMem",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "hybrid-search",
      "scope": "Agent Memory Benchmark baseline: plain hybrid search.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "32k",
      "system": "hybrid-search",
      "unit": "%",
      "value": 84.4,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": true,
      "benchmark": "LifeBench",
      "family": "answer",
      "group": "amb",
      "high": null,
      "low": null,
      "metric": "accuracy",
      "product": "hybrid-search",
      "scope": "Agent Memory Benchmark baseline: plain hybrid search.",
      "source": {
        "commit": null,
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Agent Memory Benchmark, run by Vectorize (maker of Hindsight)",
        "url": "https://agentmemorybenchmark.ai/"
      },
      "split": "en",
      "system": "hybrid-search",
      "unit": "%",
      "value": 61.0,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": 31.37,
      "low": 24.88,
      "metric": "Recall@5",
      "product": "Mnemosyne",
      "scope": "500 questions, interval 24.88 to 31.37. The scoring profile defines the recall variant; it has not been matched to the rows above, so do not compare directly.",
      "source": {
        "commit": "main",
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mnemetric LongMemEval retrieval report",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/m1.2-longmemeval-retrieval.md"
      },
      "split": "",
      "system": "Mnemosyne",
      "unit": "%",
      "value": 28.06,
      "variant": "",
      "who": "harness"
    },
    {
      "baseline": false,
      "benchmark": "LongMemEval",
      "family": "retrieval",
      "group": "",
      "high": 33.04,
      "low": 26.38,
      "metric": "nDCG@5",
      "product": "Mnemosyne",
      "scope": "500 questions, interval 26.38 to 33.04.",
      "source": {
        "commit": "main",
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mnemetric LongMemEval retrieval report",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/m1.2-longmemeval-retrieval.md"
      },
      "split": "",
      "system": "Mnemosyne",
      "unit": "%",
      "value": 29.67,
      "variant": "",
      "who": "harness"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: HotpotQA",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "Mnemosyne",
      "scope": "1,000 questions, July 2026, retrieval only.",
      "source": {
        "commit": "main",
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mnemetric HotpotQA retrieval report",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/hipporag-hotpot.md"
      },
      "split": "",
      "system": "Mnemosyne",
      "unit": "%",
      "value": 37.4,
      "variant": "",
      "who": "harness"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: 2WikiMultiHopQA",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "Mnemosyne",
      "scope": "1,000 questions, July 2026, retrieval only.",
      "source": {
        "commit": "main",
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mnemetric 2WikiMultiHopQA retrieval report",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/hipporag-2wiki.md"
      },
      "split": "",
      "system": "Mnemosyne",
      "unit": "%",
      "value": 23.725,
      "variant": "",
      "who": "harness"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: MuSiQue",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "Mnemosyne",
      "scope": "1,000 questions, July 2026, retrieval only.",
      "source": {
        "commit": "main",
        "line": null,
        "retrieved": "2026-10-06",
        "title": "Mnemetric MuSiQue retrieval report",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/hipporag-musique.md"
      },
      "split": "",
      "system": "Mnemosyne",
      "unit": "%",
      "value": 10.417,
      "variant": "",
      "who": "harness"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: HotpotQA",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "HippoRAG 2",
      "scope": "Published upstream result, Llama-3.3-70B.",
      "source": {
        "commit": "main",
        "line": 36,
        "retrieved": "2026-10-06",
        "title": "Mnemetric Phase 11 evidence note",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/phase-11-evidence.md#L36"
      },
      "split": "",
      "system": "HippoRAG 2",
      "unit": "%",
      "value": 96.3,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: 2WikiMultiHopQA",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "HippoRAG 2",
      "scope": "Published upstream result, Llama-3.3-70B.",
      "source": {
        "commit": "main",
        "line": 35,
        "retrieved": "2026-10-06",
        "title": "Mnemetric Phase 11 evidence note",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/phase-11-evidence.md#L35"
      },
      "split": "",
      "system": "HippoRAG 2",
      "unit": "%",
      "value": 90.4,
      "variant": "",
      "who": "rival"
    },
    {
      "baseline": false,
      "benchmark": "HippoRAG: MuSiQue",
      "family": "retrieval",
      "group": "",
      "high": null,
      "low": null,
      "metric": "Recall@5",
      "product": "HippoRAG 2",
      "scope": "Published upstream result, Llama-3.3-70B.",
      "source": {
        "commit": "main",
        "line": 34,
        "retrieved": "2026-10-06",
        "title": "Mnemetric Phase 11 evidence note",
        "url": "https://github.com/onfire7777/Mnemetric/blob/main/eval/reports/phase-11-evidence.md#L34"
      },
      "split": "",
      "system": "HippoRAG 2",
      "unit": "%",
      "value": 74.7,
      "variant": "",
      "who": "rival"
    }
  ],
  "corrections": [
    "Zep LongMemEval: an earlier version of the landscape page mixed the two readers in the Zep paper. The paper reports 63.8 with gpt-4o-mini (full context 55.4) and 71.2 with gpt-4o (full context 60.2).",
    "Zep latency: an earlier version used the Zep paper's latency in a chart of the Mem0 paper. The Mem0 paper measures Zep at 1.292 s p50 and 2.926 s p95 in total.",
    "LangMem latency: the figure shown was search time only (17.99 s and 59.82 s). The total is 18.53 s and 60.40 s.",
    "Removed: self-reported LoCoMo figures for supermemory (81.6) and MemOS (73.3) that could not be traced to a source. MemOS now reports 88.83; supermemory publishes a #1 claim without a figure.",
    "The capability table is an editorial summary and is labelled that way. Licences on this page come from GitHub's licence detection."
  ],
  "note": "Collected as published. Not measured by this harness unless who is harness. Families are never blended.",
  "retrieved": "2026-10-06",
  "schema_version": "mnemetric.published-claims/v1",
  "systems": [
    {
      "kind": "Open-source SDK plus managed platform",
      "licence": "Apache-2.0",
      "name": "Mem0",
      "note": "Platform scores include proprietary optimizations",
      "repository": "mem0ai/mem0"
    },
    {
      "kind": "Open source, local-first",
      "licence": "MIT",
      "name": "MemPalace",
      "note": "Declines to compare itself with other projects",
      "repository": "MemPalace/mempalace"
    },
    {
      "kind": "Open source plus cloud",
      "licence": "MIT",
      "name": "Hindsight",
      "note": "Maker of the Agent Memory Benchmark",
      "repository": "vectorize-io/hindsight"
    },
    {
      "kind": "Open-source context database",
      "licence": "AGPL-3.0",
      "name": "OpenViking",
      "note": "Reports a range, no single figure",
      "repository": "volcengine/OpenViking"
    },
    {
      "kind": "Open source",
      "licence": "MIT",
      "name": "gbrain",
      "note": "Evaluated in a separate public eval repository",
      "repository": "garrytan/gbrain"
    },
    {
      "kind": "Open source plus hosted",
      "licence": "MIT",
      "name": "supermemory",
      "note": "Claims #1 without a figure for answer quality",
      "repository": "supermemoryai/supermemory"
    },
    {
      "kind": "Open source",
      "licence": "Apache-2.0",
      "name": "agentmemory",
      "note": "Says only its own retrieval figure is measured",
      "repository": "rohitg00/agentmemory"
    },
    {
      "kind": "Memory infrastructure",
      "licence": "No standard licence detected",
      "name": "Memori",
      "note": "LoCoMo paper arXiv 2603.19935",
      "repository": "MemoriLabs/Memori"
    },
    {
      "kind": "Open-source memory OS",
      "licence": "Apache-2.0",
      "name": "MemOS",
      "note": "Scores come from its own OmniMemEval",
      "repository": "MemTensor/MemOS"
    },
    {
      "kind": "Memory layer for coding agents",
      "licence": "No standard licence detected",
      "name": "ByteRover",
      "note": "LoCoMo run on its production codebase",
      "repository": "campfirein/byterover-cli"
    },
    {
      "kind": "Personal memory layer",
      "licence": "No standard licence detected",
      "name": "CORE",
      "note": "Benchmark repo published separately",
      "repository": "RedPlanetHQ/core"
    },
    {
      "kind": "Open source",
      "licence": "MIT",
      "name": "Memanto",
      "note": "Warns scores are not comparable across projects",
      "repository": "moorcheh-ai/memanto"
    },
    {
      "kind": "Open-source knowledge graph memory",
      "licence": "Apache-2.0",
      "name": "Cognee",
      "note": "BEAM runs scoped to a few questions",
      "repository": "topoteretes/cognee"
    },
    {
      "kind": "Open-source memory kit",
      "licence": "Apache-2.0",
      "name": "ReMe",
      "note": "",
      "repository": "agentscope-ai/ReMe"
    },
    {
      "kind": "Open source",
      "licence": "MIT",
      "name": "Nemori",
      "note": "",
      "repository": "nemori-ai/nemori"
    },
    {
      "kind": "Graphiti open source; Zep Cloud managed",
      "licence": "Apache-2.0 (Graphiti)",
      "name": "Zep",
      "note": "Disputes the figure in the Mem0 paper",
      "repository": "getzep/graphiti"
    },
    {
      "kind": "Research memory framework",
      "licence": "MIT",
      "name": "LightMem",
      "note": "Its README compares several systems on one harness",
      "repository": "zjunlp/LightMem"
    },
    {
      "kind": "Open source",
      "licence": "Apache-2.0",
      "name": "Letta",
      "note": "No benchmark figures in its README",
      "repository": "letta-ai/letta"
    },
    {
      "kind": "Memory library for stateful agents",
      "licence": "AGPL-3.0",
      "name": "Honcho",
      "note": "Points to an evals page; no figures in its README",
      "repository": "plastic-labs/honcho"
    },
    {
      "kind": "Closed, built into ChatGPT",
      "licence": "Closed source",
      "name": "OpenAI memory",
      "note": "Only appears via the Mem0 paper",
      "repository": null
    },
    {
      "kind": "Operator entry of this site",
      "licence": "Not recorded here",
      "name": "Mnemosyne",
      "note": "Held to the same rules as every system",
      "repository": "onfire7777/Mnemosyne"
    }
  ]
}