{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:NMA5VANDOEABQKWDQEYLB725RZ","short_pith_number":"pith:NMA5VAND","schema_version":"1.0","canonical_sha256":"6b01da81a37100182ac38130b0ff5d8e7bc3b8ef1b03e1792c0c1d0e926abfbd","source":{"kind":"arxiv","id":"2502.15840","version":1},"attestation_state":"computed","paper":{"title":"Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Axel Backlund, Lukas Petersson","submitted_at":"2025-02-20T15:52:29Z","abstract_excerpt":"While Large Language Models (LLMs) can exhibit impressive proficiency in isolated, short-term tasks, they often fail to maintain coherent performance over longer time horizons. In this paper, we present Vending-Bench, a simulated environment designed to specifically test an LLM-based agent's ability to manage a straightforward, long-running business scenario: operating a vending machine. Agents must balance inventories, place orders, set prices, and handle daily fees - tasks that are each simple but collectively, over long horizons (>20M tokens per run) stress an LLM's capacity for sustained, "},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2502.15840","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2025-02-20T15:52:29Z","cross_cats_sorted":[],"title_canon_sha256":"adc7ea87fa723dc4d501543711fb2f7cc059a42d0cfddc0cf55ba531af81f115","abstract_canon_sha256":"ba0cc68b18019519ba1c3361d1774f9cbe00509504561baf211ea008e53b9c03"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:18:17.123888Z","signature_b64":"zPqLJ6DBAVXYGR0/oVM3RAF8pcFdMvHCyNQSKpJKQYCa0ZyHLVs98a0NZJ759IfkAltduZcKgqgFPQoG9jz9Bg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6b01da81a37100182ac38130b0ff5d8e7bc3b8ef1b03e1792c0c1d0e926abfbd","last_reissued_at":"2026-07-05T10:18:17.123401Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:18:17.123401Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Vending-Bench: A Benchmark for Long-Term Coherence of Autonomous Agents","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Axel Backlund, Lukas Petersson","submitted_at":"2025-02-20T15:52:29Z","abstract_excerpt":"While Large Language Models (LLMs) can exhibit impressive proficiency in isolated, short-term tasks, they often fail to maintain coherent performance over longer time horizons. In this paper, we present Vending-Bench, a simulated environment designed to specifically test an LLM-based agent's ability to manage a straightforward, long-running business scenario: operating a vending machine. Agents must balance inventories, place orders, set prices, and handle daily fees - tasks that are each simple but collectively, over long horizons (>20M tokens per run) stress an LLM's capacity for sustained, "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.15840","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2502.15840/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2502.15840","created_at":"2026-07-05T10:18:17.123468+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.15840v1","created_at":"2026-07-05T10:18:17.123468+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.15840","created_at":"2026-07-05T10:18:17.123468+00:00"},{"alias_kind":"pith_short_12","alias_value":"NMA5VANDOEAB","created_at":"2026-07-05T10:18:17.123468+00:00"},{"alias_kind":"pith_short_16","alias_value":"NMA5VANDOEABQKWD","created_at":"2026-07-05T10:18:17.123468+00:00"},{"alias_kind":"pith_short_8","alias_value":"NMA5VAND","created_at":"2026-07-05T10:18:17.123468+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.22909","citing_title":"Graph-Enhanced Large Language Models for Spatial Search","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19613","citing_title":"StaminaBench: Stress-Testing Coding Agents over 100 Interaction Turns","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19308","citing_title":"Enhancing Decision-Making with Large Language Models through Multi-Agent Fictitious Play","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18543","citing_title":"CEO-Bench: Can Agents Play the Long Game?","ref_index":57,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00341","citing_title":"ROGUE: Misaligned Agent Behavior Arising from Ordinary Computer Use","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2606.31916","citing_title":"Theory of Mind and Persuasion Beyond Conversation: Assessing the Capacity of LLMs to Induce Belief States via Planning and Action","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17698","citing_title":"Agent Bazaar: Enabling Economic Alignment in Multi-Agent Marketplaces","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2602.09514","citing_title":"EcoGym: Evaluating LLMs for Long-Horizon Plan-and-Execute in Interactive Economies","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.10310","citing_title":"Positive Alignment: Artificial Intelligence for Human Flourishing","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2504.19678","citing_title":"From LLM Reasoning to Autonomous AI Agents: A Comprehensive Review","ref_index":126,"is_internal_anchor":false},{"citing_arxiv_id":"2604.01444","citing_title":"Cooking Up Risks: Benchmarking and Reducing Food Safety Risks in Large Language Models","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.09678","citing_title":"NetAgentBench: A State-Centric Benchmark for Evaluating Agentic Network Configuration","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.27043","citing_title":"CL-bench Life: Can Language Models Learn from Real-Life Context?","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2602.15763","citing_title":"GLM-5: from Vibe Coding to Agentic Engineering","ref_index":3,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07733","citing_title":"CivBench: Progress-Based Evaluation for LLMs' Strategic Decision-Making in Civilization V","ref_index":8,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15267","citing_title":"CoopEval: Benchmarking Cooperation-Sustaining Mechanisms and LLM Agents in Social Dilemmas","ref_index":1,"is_internal_anchor":false},{"citing_arxiv_id":"2604.19837","citing_title":"Forage V2: Knowledge Evolution and Transfer in Autonomous Agent Organizations","ref_index":47,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ","json":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ.json","graph_json":"https://pith.science/api/pith-number/NMA5VANDOEABQKWDQEYLB725RZ/graph.json","events_json":"https://pith.science/api/pith-number/NMA5VANDOEABQKWDQEYLB725RZ/events.json","paper":"https://pith.science/paper/NMA5VAND"},"agent_actions":{"view_html":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ","download_json":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ.json","view_paper":"https://pith.science/paper/NMA5VAND","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.15840&json=true","fetch_graph":"https://pith.science/api/pith-number/NMA5VANDOEABQKWDQEYLB725RZ/graph.json","fetch_events":"https://pith.science/api/pith-number/NMA5VANDOEABQKWDQEYLB725RZ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ/action/storage_attestation","attest_author":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ/action/author_attestation","sign_citation":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ/action/citation_signature","submit_replication":"https://pith.science/pith/NMA5VANDOEABQKWDQEYLB725RZ/action/replication_record"}},"created_at":"2026-07-05T10:18:17.123468+00:00","updated_at":"2026-07-05T10:18:17.123468+00:00"}