{"work":{"id":"2698f5ad-c84c-40ca-b839-0912dae10ba2","openalex_id":null,"doi":null,"arxiv_id":"2310.08560","raw_key":null,"title":"MemGPT: Towards LLMs as Operating Systems","authors":null,"authors_text":"Charles Packer, Sarah Wooders, Kevin Lin, Vivian Fang, Shishir G. Patil, Ion Stoica","year":2023,"venue":"cs.AI","abstract":"Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers in order to effectively provide extended context within the LLM's limited context window, and utilizes interrupts to manage control flow between itself and the user. We evaluate our OS-inspired design in two domains where the limited context windows of modern LLMs severely handicaps their performance: document analysis, where MemGPT is able to analyze large documents that far exceed the underlying LLM's context window, and multi-session chat, where MemGPT can create conversational agents that remember, reflect, and evolve dynamically through long-term interactions with their users. We release MemGPT code and data for our experiments at https://memgpt.ai.","external_url":"https://arxiv.org/abs/2310.08560","cited_by_count":null,"metadata_source":"pith","metadata_fetched_at":"2026-06-29T12:43:26.043349+00:00","pith_arxiv_id":"2310.08560","created_at":"2026-05-08T17:08:34.340100+00:00","updated_at":"2026-06-29T12:43:26.043349+00:00","title_quality_ok":true,"display_title":"MemGPT: Towards LLMs as Operating Systems","render_title":"MemGPT: Towards LLMs as Operating Systems"},"hub":{"state":{"work_id":"2698f5ad-c84c-40ca-b839-0912dae10ba2","tier":"super_hub","tier_reason":"100+ Pith inbound or 10,000+ external citations","pith_inbound_count":208,"external_cited_by_count":null,"distinct_field_count":17,"first_pith_cited_at":"2023-12-18T07:47:33+00:00","last_pith_cited_at":"2026-06-25T16:36:35+00:00","author_build_status":"needed","summary_status":"needed","contexts_status":"needed","graph_status":"needed","ask_index_status":"needed","reader_status":"not_needed","recognition_status":"not_needed","updated_at":"2026-06-29T12:38:46.235237+00:00","tier_text":"super_hub"},"tier":"super_hub","role_counts":[{"context_role":"background","n":36},{"context_role":"baseline","n":3},{"context_role":"dataset","n":3},{"context_role":"method","n":1},{"context_role":"other","n":1}],"polarity_counts":[{"context_polarity":"background","n":34},{"context_polarity":"baseline","n":3},{"context_polarity":"use_dataset","n":3},{"context_polarity":"support","n":2},{"context_polarity":"unclear","n":1},{"context_polarity":"use_method","n":1}],"runs":{"ask_index":{"job_type":"ask_index","status":"succeeded","result":{"title":"MemGPT: Towards LLMs as Operating Systems","claims":[{"claim_text":"Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers i","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MemGPT: Towards LLMs as Operating Systems because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T19:16:23.399940+00:00"},"author_expand":{"job_type":"author_expand","status":"succeeded","result":{"authors_linked":[{"id":"4334cf0f-f633-4d4c-8d78-8dfadaa69abc","orcid":null,"display_name":"Charles Packer"},{"id":"605d0a01-84bb-4b24-b4fe-59136df18a11","orcid":null,"display_name":"Sarah Wooders"},{"id":"c1516ee6-d7c8-4910-97e1-a4a6b4244e6b","orcid":null,"display_name":"Kevin Lin"},{"id":"36ad8876-490d-420e-a9d8-64c3abc3e7de","orcid":null,"display_name":"Vivian Fang"},{"id":"c95fa5d3-90f2-454e-94cb-b384ae36cdb7","orcid":null,"display_name":"Shishir G. Patil"},{"id":"7e9f60ca-adad-4e13-9e27-ea1a67796d77","orcid":null,"display_name":"Ion Stoica"}]},"error":null,"updated_at":"2026-05-14T19:16:24.784950+00:00"},"context_extract":{"job_type":"context_extract","status":"succeeded","result":{"enqueued_papers":25},"error":null,"updated_at":"2026-05-14T05:46:54.141269+00:00"},"graph_features":{"job_type":"graph_features","status":"succeeded","result":{"co_cited":[{"title":"Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory","work_id":"a5aed26c-a248-48b6-a59e-f7693fcb180a","shared_citers":39},{"title":"A-MEM: Agentic Memory for LLM Agents","work_id":"3b98feb2-fdb1-479a-bbe4-2c298a4592e2","shared_citers":30},{"title":"Zep: A Temporal Knowledge Graph Architecture for Agent Memory","work_id":"515c933e-12ae-439d-a7ff-c07fee482dfb","shared_citers":23},{"title":"From Local to Global: A Graph RAG Approach to Query-Focused Summarization","work_id":"588618d7-fd41-4053-b34d-a981f8793039","shared_citers":19},{"title":"Evaluating Very Long-Term Conversational Memory of LLM Agents","work_id":"2d8c9fb7-9ace-4925-8fe1-f4e44625d04c","shared_citers":16},{"title":"Memory in the Age of AI Agents","work_id":"1ff75a14-302e-4906-aa86-1f96fbcf12ff","shared_citers":16},{"title":"Voyager: An Open-Ended Embodied Agent with Large Language Models","work_id":"ffe0d207-86cf-4742-a100-e988ac8b9676","shared_citers":16},{"title":"LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory","work_id":"9074870f-aee7-4103-b167-ac6473a8a9b3","shared_citers":15},{"title":"ReAct: Synergizing Reasoning and Acting in Language Models","work_id":"407a2351-25f1-497d-b611-f77d0292a8e6","shared_citers":15},{"title":"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation","work_id":"92b7eb9c-c3d8-4518-a376-06fa15dd895b","shared_citers":14},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":14},{"title":"Reflexion: Language Agents with Verbal Reinforcement Learning","work_id":"778f739e-5f55-4961-8a2a-e4736a2757f4","shared_citers":14},{"title":"Retrieval-Augmented Generation for Large Language Models: A Survey","work_id":"b80d2790-6cd9-4c87-b3c4-de404f99a80e","shared_citers":12},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":11},{"title":"MemoryBank: Enhancing large language models with long-term memory","work_id":"3f0e5fbe-8eb1-48c4-9e30-cf41428c3046","shared_citers":10},{"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","shared_citers":9},{"title":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks","work_id":"27eaec54-c105-4969-8188-da5f0fca3688","shared_citers":9},{"title":"Evaluating memory in llm agents via incremental multi-turn interactions","work_id":"df5096c3-e738-472b-953c-fb510ee9e68d","shared_citers":8},{"title":"Memos: A memory os for ai system","work_id":"e686f058-6c5c-4484-a88e-26d945657ed3","shared_citers":8},{"title":"title =","work_id":"b5bf85fe-4fb7-4966-b0b2-9ccf9d3b11b9","shared_citers":8},{"title":"A survey on the memory mechanism of large language model based agents","work_id":"86a3cea2-e600-466a-bf9e-6e737a41ed6a","shared_citers":7},{"title":"Generative Agents: Interactive Simulacra of Human Behavior","work_id":"01f7ddaa-284a-441a-be87-921aad4dc54b","shared_citers":7},{"title":"MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents","work_id":"64baa7f3-849b-485e-b081-2074d82f1364","shared_citers":7},{"title":"Nikita Kitaev, Lukasz Kaiser, and Anselm Levskaya","work_id":"37c05e13-4a24-44f8-a1c4-da1bbe7223aa","shared_citers":7}],"time_series":[{"n":1,"year":2024},{"n":2,"year":2025},{"n":96,"year":2026}],"dependency_candidates":[]},"error":null,"updated_at":"2026-05-14T05:56:44.441101+00:00"},"identity_refresh":{"job_type":"identity_refresh","status":"succeeded","result":{"items":[{"title":"Qwen3 Technical Report","outcome":"unchanged","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","resolver":"local_arxiv","confidence":0.98,"old_work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e"}],"counts":{"fixed":0,"merged":0,"unchanged":1,"quarantined":0,"needs_external_resolution":0},"errors":[],"attempted":1},"error":null,"updated_at":"2026-05-14T05:46:42.918872+00:00"},"role_polarity":{"job_type":"role_polarity","status":"succeeded","result":{"title":"MemGPT: Towards LLMs as Operating Systems","claims":[{"claim_text":"Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers i","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MemGPT: Towards LLMs as Operating Systems because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T19:16:23.396185+00:00"},"summary_claims":{"job_type":"summary_claims","status":"succeeded","result":{"title":"MemGPT: Towards LLMs as Operating Systems","claims":[{"claim_text":"Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers i","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MemGPT: Towards LLMs as Operating Systems because it crossed a citation-hub threshold.","role_counts":[]},"error":null,"updated_at":"2026-05-14T05:56:44.501503+00:00"}},"summary":{"title":"MemGPT: Towards LLMs as Operating Systems","claims":[{"claim_text":"Large language models (LLMs) have revolutionized AI, but are constrained by limited context windows, hindering their utility in tasks like extended conversations and document analysis. To enable using context beyond limited context windows, we propose virtual context management, a technique drawing inspiration from hierarchical memory systems in traditional operating systems that provide the appearance of large memory resources through data movement between fast and slow memory. Using this technique, we introduce MemGPT (Memory-GPT), a system that intelligently manages different memory tiers i","claim_type":"abstract","evidence_strength":"source_metadata"}],"why_cited":"Pith tracks MemGPT: Towards LLMs as Operating Systems because it crossed a citation-hub threshold.","role_counts":[]},"graph":{"co_cited":[{"title":"Mem0: Building Production-Ready AI Agents with Scalable Long-Term Memory","work_id":"a5aed26c-a248-48b6-a59e-f7693fcb180a","shared_citers":39},{"title":"A-MEM: Agentic Memory for LLM Agents","work_id":"3b98feb2-fdb1-479a-bbe4-2c298a4592e2","shared_citers":30},{"title":"Zep: A Temporal Knowledge Graph Architecture for Agent Memory","work_id":"515c933e-12ae-439d-a7ff-c07fee482dfb","shared_citers":23},{"title":"From Local to Global: A Graph RAG Approach to Query-Focused Summarization","work_id":"588618d7-fd41-4053-b34d-a981f8793039","shared_citers":19},{"title":"Evaluating Very Long-Term Conversational Memory of LLM Agents","work_id":"2d8c9fb7-9ace-4925-8fe1-f4e44625d04c","shared_citers":16},{"title":"Memory in the Age of AI Agents","work_id":"1ff75a14-302e-4906-aa86-1f96fbcf12ff","shared_citers":16},{"title":"Voyager: An Open-Ended Embodied Agent with Large Language Models","work_id":"ffe0d207-86cf-4742-a100-e988ac8b9676","shared_citers":16},{"title":"LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory","work_id":"9074870f-aee7-4103-b167-ac6473a8a9b3","shared_citers":15},{"title":"ReAct: Synergizing Reasoning and Acting in Language Models","work_id":"407a2351-25f1-497d-b611-f77d0292a8e6","shared_citers":15},{"title":"AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation","work_id":"92b7eb9c-c3d8-4518-a376-06fa15dd895b","shared_citers":14},{"title":"Qwen3 Technical Report","work_id":"25a4e30c-1232-48e7-9925-02fa12ba7c9e","shared_citers":14},{"title":"Reflexion: Language Agents with Verbal Reinforcement Learning","work_id":"778f739e-5f55-4961-8a2a-e4736a2757f4","shared_citers":14},{"title":"Retrieval-Augmented Generation for Large Language Models: A Survey","work_id":"b80d2790-6cd9-4c87-b3c4-de404f99a80e","shared_citers":12},{"title":"GPT-4 Technical Report","work_id":"b928e041-6991-4c08-8c81-0359e4097c7b","shared_citers":11},{"title":"MemoryBank: Enhancing large language models with long-term memory","work_id":"3f0e5fbe-8eb1-48c4-9e30-cf41428c3046","shared_citers":10},{"title":"Evaluating Large Language Models Trained on Code","work_id":"042493e9-b26f-4b4e-bbde-382072ca9b08","shared_citers":9},{"title":"Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks","work_id":"27eaec54-c105-4969-8188-da5f0fca3688","shared_citers":9},{"title":"Evaluating memory in llm agents via incremental multi-turn interactions","work_id":"df5096c3-e738-472b-953c-fb510ee9e68d","shared_citers":8},{"title":"Memos: A memory os for ai system","work_id":"e686f058-6c5c-4484-a88e-26d945657ed3","shared_citers":8},{"title":"title =","work_id":"b5bf85fe-4fb7-4966-b0b2-9ccf9d3b11b9","shared_citers":8},{"title":"A survey on the memory mechanism of large language model based agents","work_id":"86a3cea2-e600-466a-bf9e-6e737a41ed6a","shared_citers":7},{"title":"Generative Agents: Interactive Simulacra of Human Behavior","work_id":"01f7ddaa-284a-441a-be87-921aad4dc54b","shared_citers":7},{"title":"MemSkill: Learning and Evolving Memory Skills for Self-Evolving Agents","work_id":"64baa7f3-849b-485e-b081-2074d82f1364","shared_citers":7},{"title":"Nikita Kitaev, Lukasz Kaiser, and Anselm Levskaya","work_id":"37c05e13-4a24-44f8-a1c4-da1bbe7223aa","shared_citers":7}],"time_series":[{"n":1,"year":2024},{"n":2,"year":2025},{"n":96,"year":2026}],"dependency_candidates":[]},"authors":[{"id":"4334cf0f-f633-4d4c-8d78-8dfadaa69abc","orcid":null,"display_name":"Charles Packer","source":"manual","import_confidence":0.72},{"id":"7e9f60ca-adad-4e13-9e27-ea1a67796d77","orcid":null,"display_name":"Ion Stoica","source":"manual","import_confidence":0.72},{"id":"c1516ee6-d7c8-4910-97e1-a4a6b4244e6b","orcid":null,"display_name":"Kevin Lin","source":"manual","import_confidence":0.72},{"id":"605d0a01-84bb-4b24-b4fe-59136df18a11","orcid":null,"display_name":"Sarah Wooders","source":"manual","import_confidence":0.72},{"id":"c95fa5d3-90f2-454e-94cb-b384ae36cdb7","orcid":null,"display_name":"Shishir G. Patil","source":"manual","import_confidence":0.72},{"id":"36ad8876-490d-420e-a9d8-64c3abc3e7de","orcid":null,"display_name":"Vivian Fang","source":"manual","import_confidence":0.72}]}}