Pith Number

pith:PFSVRTVM

pith:2026:PFSVRTVMWP6CFX6LFLP5JHUH54

not attested not anchored not stored refs resolved

GroupMemBench: Benchmarking LLM Agent Memory in Multi-Party Conversations

Evgeniy Gabrilovich, Jingbo Yang, Kwei-Herng Lai, Shiyu Chang, Xiaowen Wang, Yaar Harari

Benchmarking shows leading LLM memory systems reach only 46 percent accuracy in multi-party conversations, with BM25 matching most.

arxiv:2605.14498 v1 · 2026-05-14 · cs.CL

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{PFSVRTVMWP6CFX6LFLP5JHUH54}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Benchmarking leading memory systems exposes a sharp collapse: the strongest one reaches only 46.0% average accuracy, with knowledge update at 27.1% and term ambiguity at 37.7%, while a simple BM25 baseline matches or exceeds most agent memory systems.

C2weakest assumption

The graph-grounded synthesis pipeline and adversarial query generation produce conversations and questions that faithfully capture the three unmeasured properties of group memory (group dynamics, speaker-grounded belief tracking, and audience-adapted language) in real deployments.

C3one line summary

GroupMemBench shows leading LLM memory systems reach only 46% average accuracy on multi-party tasks, with a simple BM25 baseline matching or beating most of them.

References

47 extracted · 47 resolved · 10 Pith anchors

[1] OpenClaw-RL: Train Any Agent Simply by Talking 2026 · arXiv:2603.10165

[2] Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward 2026 · arXiv:2602.12430

[3] Trust in ai chatbots: A systematic review.Telematics and Informatics, 97:102240, 2025 2025

[4] Personal LLM Agents: Insights and Survey about the Capability, Efficiency and Security 2024 · arXiv:2401.05459

[5] A survey on large language model based autonomous agents.Frontiers of Computer Science, 18(6):186345 2024

Receipt and verification

First computed	2026-05-17T23:39:06.349356Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

796558ceacb3fc22dfcb2adfd49e87ef0737163bb18c20c2945ae60cee375576

Aliases

arxiv: 2605.14498 · arxiv_version: 2605.14498v1 · doi: 10.48550/arxiv.2605.14498 · pith_short_12: PFSVRTVMWP6C · pith_short_16: PFSVRTVMWP6CFX6L · pith_short_8: PFSVRTVM

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/PFSVRTVMWP6CFX6LFLP5JHUH54 \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 796558ceacb3fc22dfcb2adfd49e87ef0737163bb18c20c2945ae60cee375576

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "88edf3ae25000759317ae305503a4c09b7cc901c943ecec0d81cdee0e80a20e2",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-14T07:38:29Z",
    "title_canon_sha256": "2cb1a9babb0c11bd583a0fb06ade13e53199b2773ceb18a9971d09d21f9de78f"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.14498",
    "kind": "arxiv",
    "version": 1
  }
}