Structured registry of agent memory benchmarks, with every field traced to its source
HTML
173
25 commits
updated Oct 12, 2026
English | 中文
Browse the registry online → Search and filter by memory type, reproducibility level and year.
A structured registry of agent memory benchmarks: one YAML file per benchmark (benchmarks/<slug>.yaml). This README is generated by a script.
Every field must be traceable to the paper or the official repository; where each value came from is recorded in the entry's sources. Anything that cannot be found is null.
207 entries. Reproducibility levels: unverified 0 · paper_only 37 · data_public 31 · official_script 139 · smoke_tested 0 · full_run 0
A benchmark can belong to several types. Expand a type to see its entries, newest first.
| Type | What the benchmark tests |
|---|---|
conversational_longterm | Recall or reasoning over long, multi-session dialogue or chat histories (e.g. LoCoMo, LongMemEval, MSC) |
personalization | Using user-specific preferences, personas or profiles to tailor answers (e.g. PersonaMem, PrefEval) |
episodic | Recalling specific events and experiences with time, place or order ("what happened when") |
semantic_knowledge_update | Factual knowledge stored outside the weights (context or external store), updated correctly when facts change |
procedural_workflow | Distilling skills, workflows or tool use from earlier tasks and reusing them on different later tasks |
coding_cross_session | Coding agents that must retain information across independent sessions or tasks |
parametric | Memory in the model weights: knowledge editing, injection by fine-tuning, test-time training, memorization or unlearning |
multi_agent_group | Memory shared among several agents, or multi-party / group conversations |
multimodal | Remembered content includes non-text input (images, video, audio) that the system must process directly |
embodied | Agents act in a simulated or real environment; memory is scored through actions or task success |
memory_security | Attacks on or protection of memory: poisoning, injection, privacy leakage, whether deletion requests take effect |
Single-input long-context tests (e.g. RULER, NIAH, LongBench) are not memory benchmarks and are not included.
conversational_longterm (81)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| LoCoMo-Conv | 2026‑09 | paper · code · data | 10 conv. 6,581 Q | official_script |
| LSREP | 2026‑09 | paper · code | 4 conv. 219 Q | paper_only |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| TWIST | 2026‑09 | paper · code · data | 161 Q | official_script |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MemUse | 2026‑08 | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| StateMemBench | 2026‑08 | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 2026‑08 | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| InMind | 2026‑07 | paper · code | 1 conv. 125 Q | data_public |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 2026‑06 | paper · code | 20 conv. 15,422 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| MemAudit | 2026‑05 | paper · code · data | 87 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| RefMem-Bench | 2026‑05 | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| StratMem-Bench | 2026‑04 | paper · code | 657 Q | data_public |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 2026‑03 | paper · code | 100 conv. 1,300 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 2026‑02 | paper | — | paper_only |
| LoCoMo-Plus | 2026‑02 | paper · code | 10 conv. 401 Q | official_script |
| Neuromem | 2026‑02 | paper · code | 10 conv. | official_script |
| CAME-Bench | 2026‑01 | paper · code · data | 14 conv. 373 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| MemoryRewardBench | 2026‑01 | paper · code · data | 2,400 Q | official_script |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemoryBench | 2025‑10 | paper · code · data | 4,063 Q | official_script |
| ZH-4O | 2025‑09 | paper · code | 28 conv. 1,068 Q | official_script |
| OdysseyBench | 2025‑08 | paper · code | 602 conv. 602 Q | official_script |
| LOCCO | 2025‑07 | paper · code | 100 conv. 2,981 Q | data_public |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| MS-TOD | 2025‑05 | paper · code | 132 conv. 956 Q | official_script |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| ImplexConv | 2025‑03 | paper · code · data | 2,500 conv. | data_public |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 2025‑02 | paper · code · data | 3,000 Q | official_script |
| REALTALK | 2025‑02 | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 2025‑02 | paper | 600 Q | paper_only |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 2024‑09 | paper · code | 160 conv. | official_script |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
| TeaFarm | 2024‑06 | paper · code · data | 100 conv. 200 Q | data_public |
| LoCoMo | 2024‑02 | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| DMR (Deep Memory Retrieval) | 2023‑10 | paper · code · data | 500 conv. 500 Q | data_public |
| MemoryBank | 2023‑05 | paper · code | 15 conv. 194 Q | data_public |
| MemoEval (SCM Self-Controlled Memory evaluation dataset) | 2023‑04 | paper · code | 18 conv. 105 Q | data_public |
| Multi-Session Chat (MSC) | 2021‑07 | paper · code · data | 501 conv. | official_script |
personalization (68)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| CMSC-E | 2026‑10 | paper · code · data | — | official_script |
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| Lapse | 2026‑09 | paper | 128 Q | paper_only |
| MemCalib | 2026‑09 | paper · code · data | 1,500 Q | data_public |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| PairPref | 2026‑09 | paper | 1,227 Q | paper_only |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| MemUse | 2026‑08 | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 2026‑08 | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| CAPA | 2026‑07 | paper · code | 60 conv. 300 Q | data_public |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| InMind | 2026‑07 | paper · code | 1 conv. 125 Q | data_public |
| MemSyco-Bench | 2026‑07 | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 2026‑07 | paper · code | — | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MIST | 2026‑06 | paper | 598 conv. 598 Q | paper_only |
| StreamMemBench | 2026‑06 | paper · code | 16,214 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| RefMem-Bench | 2026‑05 | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 2026‑03 | paper · code | 100 conv. 1,300 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| LoCoMo-Plus | 2026‑02 | paper · code | 10 conv. 401 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| PersistBench | 2026‑02 | paper · code | 500 Q | official_script |
| AndroidIntent | 2026‑01 | paper · code · data | 91 conv. 990 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| KnowMe-Bench | 2026‑01 | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| RPEval | 2026‑01 | paper · code | 953 Q | official_script |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| BELA | 2025‑11 | paper · code · data | 40 conv. | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| Stateful SWE-bench | 2025‑10 | paper · code · data | 15 conv. 500 Q | official_script |
| ZH-4O | 2025‑09 | paper · code | 28 conv. 1,068 Q | official_script |
| MPR | 2025‑08 | paper · code · data | 12 conv. 108,000 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| ImplexConv | 2025‑03 | paper · code · data | 2,500 conv. | data_public |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 2025‑02 | paper · code · data | 3,000 Q | official_script |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| MemDaily | 2024‑09 | paper · code | 2,954 conv. 2,954 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| Multi-Session Chat (MSC) | 2021‑07 | paper · code · data | 501 conv. | official_script |
episodic (56)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| ICM-Bench | 2026‑09 | paper · code · data | 839 conv. 1,217 Q | official_script |
| MemTransfer | 2026‑09 | paper | 100 conv. 100 Q | paper_only |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| AgenticSTS | 2026‑07 | paper · code · data | — | official_script |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| MERIT | 2026‑07 | paper · code | — | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| DMV-Bench | 2026‑06 | paper · code · data | 35,410 Q | official_script |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| StreamMemBench | 2026‑06 | paper · code | 16,214 Q | official_script |
| EgoStream | 2026‑05 | paper | 2,250 Q | paper_only |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SuperMemory-VQA | 2026‑05 | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| AMA-Bench | 2026‑02 | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| KnowMe-Bench | 2026‑01 | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemTrack | 2025‑10 | paper · data | 47 conv. | data_public |
| TeleEgo | 2025‑10 | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 2025‑09 | paper · code | 34 Q | official_script |
| M3-Bench | 2025‑08 | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| StuLife | 2025‑08 | paper · code · data | 1,284 Q | official_script |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| StoryBench | 2025‑06 | paper | 86 Q | paper_only |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| REALTALK | 2025‑02 | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 2025‑02 | paper | 600 Q | paper_only |
| Episodic Memory Benchmark | 2025‑01 | paper · code · data | 686 Q | official_script |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 2024‑09 | paper · code | 160 conv. | official_script |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
| LoCoMo | 2024‑02 | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| MemoryBank | 2023‑05 | paper · code | 15 conv. 194 Q | data_public |
semantic_knowledge_update (74)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| Knowing the Store | 2026‑10 | paper · code · data | 444 Q | official_script |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| Controlled In-Context Memory | 2026‑09 | paper | 1,200 conv. 1,200 Q | paper_only |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| EvoPathBench | 2026‑09 | paper · code | 1,800 conv. 3,600 Q | official_script |
| Impact Is Not Invalidation | 2026‑09 | paper | 10,369 Q | paper_only |
| Lapse | 2026‑09 | paper | 128 Q | paper_only |
| LSREP | 2026‑09 | paper · code | 4 conv. 219 Q | paper_only |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| MemRiskBench | 2026‑09 | paper · code · data | 120 conv. 120 Q | official_script |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| agent-memory-bench | 2026‑08 | code · data | 196 conv. 26 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| StateMemBench | 2026‑08 | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MemSyco-Bench | 2026‑07 | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 2026‑07 | paper · code | — | official_script |
| PM-Bench | 2026‑07 | paper · code | 1 conv. 81 Q | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| ForgetEval | 2026‑06 | paper · code | 385 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 2026‑06 | paper · code | 20 conv. 15,422 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| MemAudit | 2026‑05 | paper · code · data | 87 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| PrecisionMemBench | 2026‑05 | paper · code · data | 2 conv. 89 Q | official_script |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| AMA-Bench | 2026‑02 | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 2026‑02 | paper | — | paper_only |
| Neuromem | 2026‑02 | paper · code | 10 conv. | official_script |
| CAME-Bench | 2026‑01 | paper · code · data | 14 conv. 373 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemTrack | 2025‑10 | paper · data | 47 conv. | data_public |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
| Minerva | 2025‑02 | paper · code | 1,110 Q | official_script |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| EvoWiki | 2024‑12 | paper · code · data | 10,264 Q | paper_only |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MQuAKE | 2023‑05 | paper · code | 3,000 Q | official_script |
procedural_workflow (53)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| Learn2Play Bench | 2026‑10 | paper · code | 20 Q | data_public |
| MetaKernelBench | 2026‑10 | paper · code | 74 Q | official_script |
| MLE-InsightBench | 2026‑10 | paper | 12 Q | paper_only |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| ConsolidationBench | 2026‑09 | paper | — | paper_only |
| EconSkills | 2026‑09 | paper · data | 360 Q | data_public |
| EngramBench | 2026‑09 | paper · code | 30 conv. 13 Q | official_script |
| EvoHarnessBench | 2026‑09 | paper · data | 802 Q | data_public |
| EvoPathBench | 2026‑09 | paper · code | 1,800 conv. 3,600 Q | official_script |
| VibeMemBench | 2026‑09 | paper · code | 3,634 conv. 111 Q | paper_only |
| ContinualSkillBench | 2026‑08 | paper · code | 500 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| FinEvo-Bench | 2026‑08 | paper | 120 Q | paper_only |
| GDPevo | 2026‑08 | paper · code | 240 Q | official_script |
| MemTrapBench | 2026‑08 | paper · code | 1,050 conv. 1,050 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| PATH-Bench | 2026‑08 | paper | 120 Q | paper_only |
| S3Gym | 2026‑08 | paper | — | paper_only |
| AgenticSTS | 2026‑07 | paper · code · data | — | official_script |
| AgentStream | 2026‑07 | paper · code | 300 Q | official_script |
| EvoAgentBench | 2026‑07 | paper · code · data | 267 Q | official_script |
| SLVMBench | 2026‑07 | paper · data | 2,261 Q | data_public |
| AFTER | 2026‑06 | paper · code · data | 382 Q | data_public |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| AhaBench | 2026‑06 | paper · code | — | official_script |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| FinEvolveBench | 2026‑06 | paper · code · data | — | official_script |
| MedEvoEval | 2026‑06 | paper | 700 Q | paper_only |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| SEAGym | 2026‑06 | paper · code | 55 Q | official_script |
| BenchTrace | 2026‑05 | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| SeqMem-Eval | 2026‑05 | paper · code | 3,726 Q | official_script |
| SkillEvolBench | 2026‑05 | paper · code · data | 180 Q | official_script |
| STATE-Bench | 2026‑05 | code | 300 conv. 150 Q | official_script |
| SEA-Eval | 2026‑04 | paper · code | 92 conv. 32 Q | data_public |
| SkillFlow | 2026‑04 | paper · code · data | 166 Q | official_script |
| SkillLearnBench | 2026‑04 | paper · code | 100 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| SkillCraft | 2026‑02 | paper · code | 126 Q | official_script |
| SWE-ContextBench | 2026‑02 | paper · code · data | 376 Q | official_script |
| Trust-Memevo | 2026‑02 | paper · code | 2,329 Q | paper_only |
| TraineeBench | 2026‑01 | paper · code | 50 conv. 203 Q | official_script |
| BELA | 2025‑11 | paper · code · data | 40 conv. | official_script |
| Evo-Memory | 2025‑11 | paper | 3,716 Q | paper_only |
| MemoryBench | 2025‑10 | paper · code · data | 4,063 Q | official_script |
| StuLife | 2025‑08 | paper · code · data | 1,284 Q | official_script |
| EvaLearn | 2025‑06 | paper · code | 182 conv. 648 Q | official_script |
| SWE-Bench-CL | 2025‑06 | paper · code · data | 8 conv. 273 Q | official_script |
| LifelongAgentBench | 2025‑05 | paper · code · data | 1,396 Q | official_script |
| StreamBench | 2024‑06 | paper · code · data | 9,702 Q | official_script |
coding_cross_session (17)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| MetaKernelBench | 2026‑10 | paper · code | 74 Q | official_script |
| EngramBench | 2026‑09 | paper · code | 30 conv. 13 Q | official_script |
| VibeMemBench | 2026‑09 | paper · code | 3,634 conv. 111 Q | paper_only |
| agent-memory-bench | 2026‑08 | code · data | 196 conv. 26 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| CAPA | 2026‑07 | paper · code | 60 conv. 300 Q | data_public |
| ChainSWE | 2026‑07 | paper · data | 100 conv. 304 Q | data_public |
| EvoAgentBench | 2026‑07 | paper · code · data | 267 Q | official_script |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 2026‑07 | paper · code | 194 Q | official_script |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| SWE-ContextBench | 2026‑02 | paper · code · data | 376 Q | official_script |
| Stateful SWE-bench | 2025‑10 | paper · code · data | 15 conv. 500 Q | official_script |
| SWE-Bench-CL | 2025‑06 | paper · code · data | 8 conv. 273 Q | official_script |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
parametric (27)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| Cross-lingual Unlearning Tensor | 2026‑09 | paper · code | 4,000 Q | paper_only |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| MemoReason | 2026‑09 | paper · code · data | 1,200 Q | official_script |
| MemToC | 2026‑08 | paper · code | 6,504 Q | data_public |
| S3Gym | 2026‑08 | paper | — | paper_only |
| ForgetBench | 2026‑07 | paper | 6,431 Q | paper_only |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MPR | 2025‑08 | paper · code · data | 12 conv. 108,000 Q | official_script |
| LOCCO | 2025‑07 | paper · code | 100 conv. 2,981 Q | data_public |
| OpenUnlearning | 2025‑06 | paper · code · data | 400 Q | official_script |
| UniEdit | 2025‑05 | paper · code · data | 311,142 Q | official_script |
| WikiDYK | 2025‑05 | paper · code · data | 77,180 Q | official_script |
| KUP (Knowledge Update Playground) | 2025‑04 | paper · code · data | 6,260 Q | official_script |
| TiC-LM | 2025‑04 | paper · code | — | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| WikiBigEdit | 2025‑03 | paper · code · data | 502,382 Q | official_script |
| QAEdit | 2025‑02 | paper · code · data | 19,249 Q | official_script |
| Episodic Memory Benchmark | 2025‑01 | paper · code · data | 686 Q | official_script |
| EvoWiki | 2024‑12 | paper · code · data | 10,264 Q | paper_only |
| HalluEditBench | 2024‑10 | paper · code · data | 6,738 Q | official_script |
| CodeUpdateArena | 2024‑07 | paper · code · data | 670 Q | official_script |
| KnowEdit | 2024‑01 | paper · code · data | 5,653 Q | official_script |
| AToKe | 2023‑12 | paper · code | 8,820 Q | data_public |
| ConflictEdit / RoundEdit | 2023‑10 | paper · code | 2,500 Q | official_script |
| MMEdit | 2023‑10 | paper · code · data | 3,093 Q | official_script |
| RippleEdits | 2023‑07 | paper · code | 4,000 Q | official_script |
| MQuAKE | 2023‑05 | paper · code | 3,000 Q | official_script |
multi_agent_group (8)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| CoMemBench | 2026‑09 | paper | 800 Q | paper_only |
| Correlated Promotion Benchmark (CPB) | 2026‑09 | paper · code · data | 100 conv. | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
multimodal (28)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| ICM-Bench | 2026‑09 | paper · code · data | 839 conv. 1,217 Q | official_script |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| Video2Skill | 2026‑09 | paper · data | 172 conv. 1,671 Q | data_public |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| Sequential-EQA | 2026‑07 | paper · code | 50 conv. 498 Q | official_script |
| SLVMBench | 2026‑07 | paper · data | 2,261 Q | data_public |
| DMV-Bench | 2026‑06 | paper · code · data | 35,410 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| EgoStream | 2026‑05 | paper | 2,250 Q | paper_only |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| MBench | 2026‑05 | paper · code · data | 1,040 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SuperMemory-VQA | 2026‑05 | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RMBench | 2026‑03 | paper · code · data | — | official_script |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| TeleEgo | 2025‑10 | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 2025‑09 | paper · code | 34 Q | official_script |
| M3-Bench | 2025‑08 | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| MMEdit | 2023‑10 | paper · code · data | 3,093 Q | official_script |
embodied (12)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| MemTransfer | 2026‑09 | paper | 100 conv. 100 Q | paper_only |
| Sequential-EQA | 2026‑07 | paper · code | 50 conv. 498 Q | official_script |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| BenchTrace | 2026‑05 | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SeqMem-Eval | 2026‑05 | paper · code | 3,726 Q | official_script |
| RMBench | 2026‑03 | paper · code · data | — | official_script |
| Evo-Memory | 2025‑11 | paper | 3,716 Q | paper_only |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
memory_security (18)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| ConsolidationBench | 2026‑09 | paper | — | paper_only |
| Correlated Promotion Benchmark (CPB) | 2026‑09 | paper · code · data | 100 conv. | official_script |
| Cross-lingual Unlearning Tensor | 2026‑09 | paper · code | 4,000 Q | paper_only |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| MemRiskBench | 2026‑09 | paper · code · data | 120 conv. 120 Q | official_script |
| TWIST | 2026‑09 | paper · code · data | 161 Q | official_script |
| MemTrapBench | 2026‑08 | paper · code | 1,050 conv. 1,050 Q | official_script |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 2026‑07 | paper · code | 194 Q | official_script |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MemSecBench | 2026‑07 | paper | 310 Q | paper_only |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| ForgetEval | 2026‑06 | paper · code | 385 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MPBench | 2026‑06 | paper · code | 3,240 Q | data_public |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| OpenUnlearning | 2025‑06 | paper · code · data | 400 Q | official_script |
Grouped by year, newest first. Click a name to see all fields and their sources.
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| CMSC-E | 10 | personalization | paper · code · data | — | official_script |
| DyadMem | 10 | conversational_longterm, personalization, episodic, semantic_knowledge_update | paper | 3,065 conv. 61,210 Q | paper_only |
| Knowing the Store | 10 | semantic_knowledge_update | paper · code · data | 444 Q | official_script |
| Learn2Play Bench | 10 | procedural_workflow | paper · code | 20 Q | data_public |
| MetaKernelBench | 10 | procedural_workflow, coding_cross_session | paper · code | 74 Q | official_script |
| MLE-InsightBench | 10 | procedural_workflow | paper | 12 Q | paper_only |
| RealCompanion | 10 | conversational_longterm, personalization, episodic | paper · code · data | 10 conv. 3,235 Q | paper_only |
| ServeLearnBench | 10 | procedural_workflow, semantic_knowledge_update, personalization | paper · code | 3,210 Q | official_script |
| APM-Bench | 09 | episodic, semantic_knowledge_update, multimodal | paper · code · data | 104 conv. 2,719 Q | paper_only |
| CoMemBench | 09 | multi_agent_group | paper | 800 Q | paper_only |
| ConsolidationBench | 09 | procedural_workflow, memory_security | paper | — | paper_only |
| Controlled In-Context Memory | 09 | semantic_knowledge_update | paper | 1,200 conv. 1,200 Q | paper_only |
| Correlated Promotion Benchmark (CPB) | 09 | multi_agent_group, memory_security | paper · code · data | 100 conv. | official_script |
| Cross-lingual Unlearning Tensor | 09 | parametric, memory_security | paper · code | 4,000 Q | paper_only |
| CUE-Mem | 09 | conversational_longterm, personalization, multimodal | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 09 | conversational_longterm, personalization, episodic | paper · code · data | 3 conv. 600 Q | official_script |
| EconSkills | 09 | procedural_workflow | paper · data | 360 Q | data_public |
| EmbodiedMemory-Bench (EMem-Bench) | 09 | embodied, episodic, semantic_knowledge_update, multimodal | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| EngramBench | 09 | procedural_workflow, coding_cross_session | paper · code | 30 conv. 13 Q | official_script |
| EvoHarnessBench | 09 | procedural_workflow | paper · data | 802 Q | data_public |
| EvoPathBench | 09 | procedural_workflow, semantic_knowledge_update | paper · code | 1,800 conv. 3,600 Q | official_script |
| ICM-Bench | 09 | episodic, multimodal | paper · code · data | 839 conv. 1,217 Q | official_script |
| IDUnlearn-Bench | 09 | parametric, memory_security, multimodal | paper · code · data | 4,860 Q | data_public |
| Impact Is Not Invalidation | 09 | semantic_knowledge_update | paper | 10,369 Q | paper_only |
| Lapse | 09 | personalization, semantic_knowledge_update | paper | 128 Q | paper_only |
| LoCoMo-Conv | 09 | conversational_longterm | paper · code · data | 10 conv. 6,581 Q | official_script |
| LSREP | 09 | conversational_longterm, semantic_knowledge_update | paper · code | 4 conv. 219 Q | paper_only |
| MemCalib | 09 | personalization | paper · code · data | 1,500 Q | data_public |
| MemoReason | 09 | parametric | paper · code · data | 1,200 Q | official_script |
| MemProbe | 09 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 56 conv. 224 Q | official_script |
| MemRiskBench | 09 | semantic_knowledge_update, memory_security | paper · code · data | 120 conv. 120 Q | official_script |
| MemTransfer | 09 | episodic, embodied | paper | 100 conv. 100 Q | paper_only |
| PairPref | 09 | personalization | paper | 1,227 Q | paper_only |
| PRAGMA | 09 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 100 conv. 400 Q | data_public |
| TWIST | 09 | conversational_longterm, memory_security | paper · code · data | 161 Q | official_script |
| VibeMemBench | 09 | coding_cross_session, procedural_workflow | paper · code | 3,634 conv. 111 Q | paper_only |
| Video2Skill | 09 | multimodal | paper · data | 172 conv. 1,671 Q | data_public |
| VoxMem | 09 | conversational_longterm, multimodal, semantic_knowledge_update, episodic | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| agent-memory-bench | 08 | coding_cross_session, semantic_knowledge_update | code · data | 196 conv. 26 Q | official_script |
| ContinualSkillBench | 08 | procedural_workflow | paper · code | 500 Q | official_script |
| DreamBench-SWE | 08 | coding_cross_session, semantic_knowledge_update, procedural_workflow | paper · code | 60 conv. 60 Q | data_public |
| FinEvo-Bench | 08 | procedural_workflow | paper | 120 Q | paper_only |
| GDPevo | 08 | procedural_workflow | paper · code | 240 Q | official_script |
| MemToC | 08 | parametric | paper · code | 6,504 Q | data_public |
| MemTrapBench | 08 | procedural_workflow, memory_security | paper · code | 1,050 conv. 1,050 Q | official_script |
| MemUse | 08 | conversational_longterm, personalization | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 08 | conversational_longterm, personalization, episodic, semantic_knowledge_update, multimodal | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 08 | procedural_workflow, semantic_knowledge_update, personalization, episodic | paper · code | 26 conv. 204 Q | official_script |
| PATH-Bench | 08 | procedural_workflow | paper | 120 Q | paper_only |
| S3Gym | 08 | procedural_workflow, parametric | paper | — | paper_only |
| StateMemBench | 08 | conversational_longterm, semantic_knowledge_update | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 08 | personalization, conversational_longterm, semantic_knowledge_update | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 08 | conversational_longterm, personalization | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| AgenticSTS | 07 | procedural_workflow, episodic | paper · code · data | — | official_script |
| AgentStream | 07 | procedural_workflow | paper · code | 300 Q | official_script |
| CAPA | 07 | coding_cross_session, personalization | paper · code | 60 conv. 300 Q | data_public |
| ChainSWE | 07 | coding_cross_session | paper · data | 100 conv. 304 Q | data_public |
| EvoAgentBench | 07 | procedural_workflow, coding_cross_session | paper · code · data | 267 Q | official_script |
| ForgetBench | 07 | parametric | paper | 6,431 Q | paper_only |
| Ground Truth First | 07 | conversational_longterm, personalization, episodic, semantic_knowledge_update, memory_security | paper · code | 383 Q | paper_only |
| InMind | 07 | personalization, conversational_longterm | paper · code | 1 conv. 125 Q | data_public |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 07 | coding_cross_session, memory_security | paper · code | 194 Q | official_script |
| MemOps | 07 | conversational_longterm, semantic_knowledge_update, parametric, memory_security | paper · code | 403 conv. 2,006 Q | official_script |
| MemSecBench | 07 | memory_security | paper | 310 Q | paper_only |
| MemSyco-Bench | 07 | personalization, semantic_knowledge_update | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 07 | episodic, semantic_knowledge_update, personalization | paper · code | — | official_script |
| PM-Bench | 07 | semantic_knowledge_update | paper · code | 1 conv. 81 Q | official_script |
| RUMBA | 07 | conversational_longterm, episodic, semantic_knowledge_update, memory_security | paper · code · data | 85 conv. 1,543 Q | official_script |
| Sequential-EQA | 07 | embodied, multimodal | paper · code | 50 conv. 498 Q | official_script |
| SLVMBench | 07 | procedural_workflow, multimodal | paper · data | 2,261 Q | data_public |
| AFTER | 06 | procedural_workflow | paper · code · data | 382 Q | data_public |
| AgentCL | 06 | procedural_workflow, coding_cross_session, embodied | paper · data | — | data_public |
| AhaBench | 06 | procedural_workflow | paper · code | — | official_script |
| Continual Learning Bench (CL-Bench) | 06 | procedural_workflow, coding_cross_session, semantic_knowledge_update | paper · code · data | 6 conv. 301 Q | official_script |
| DMV-Bench | 06 | episodic, multimodal | paper · code · data | 35,410 Q | official_script |
| EvoArena | 06 | semantic_knowledge_update, procedural_workflow, coding_cross_session, personalization, conversational_longterm | paper · code · data | 10 conv. 505 Q | official_script |
| FinEvolveBench | 06 | procedural_workflow | paper · code · data | — | official_script |
| ForgetEval | 06 | semantic_knowledge_update, memory_security | paper · code | 385 Q | official_script |
| GateMem | 06 | multi_agent_group, memory_security, semantic_knowledge_update, conversational_longterm | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 06 | conversational_longterm, multimodal, multi_agent_group, semantic_knowledge_update | paper · code · data | 25 conv. 2,236 Q | official_script |
| HUSH-Bench | 06 | memory_security, personalization, conversational_longterm | paper | 10 conv. 2,400 Q | paper_only |
| MedEvoEval | 06 | procedural_workflow | paper | 700 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 06 | personalization, conversational_longterm, episodic | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 06 | conversational_longterm, multi_agent_group, semantic_knowledge_update, episodic | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 06 | conversational_longterm, semantic_knowledge_update | paper · code | 20 conv. 15,422 Q | official_script |
| MIST | 06 | personalization | paper | 598 conv. 598 Q | paper_only |
| MPBench | 06 | memory_security | paper · code | 3,240 Q | data_public |
| OmniGameArena | 06 | procedural_workflow, embodied, multimodal | paper · code · data | — | official_script |
| SEAGym | 06 | procedural_workflow | paper · code | 55 Q | official_script |
| StreamMemBench | 06 | personalization, episodic | paper · code | 16,214 Q | official_script |
| SubtleMemory | 06 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 10 conv. 1,522 Q | official_script |
| BenchTrace | 05 | procedural_workflow, embodied | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EgoStream | 05 | episodic, multimodal | paper | 2,250 Q | paper_only |
| EvoMemBench | 05 | procedural_workflow, semantic_knowledge_update, embodied, conversational_longterm | paper · code | 5,754 Q | official_script |
| GroupMemBench | 05 | multi_agent_group, conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code · data | 4 conv. 745 Q | official_script |
| LongMemEval-V2 | 05 | episodic, procedural_workflow, multimodal | paper · code · data | 1,540 conv. 451 Q | official_script |
| MBench | 05 | multimodal | paper · code · data | 1,040 Q | official_script |
| MemAudit | 05 | semantic_knowledge_update, conversational_longterm | paper · code · data | 87 Q | official_script |
| MemEye | 05 | conversational_longterm, multimodal, semantic_knowledge_update | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 05 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 161 Q | official_script |
| PrecisionMemBench | 05 | semantic_knowledge_update | paper · code · data | 2 conv. 89 Q | official_script |
| RefMem-Bench | 05 | conversational_longterm, personalization | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 05 | conversational_longterm, episodic, semantic_knowledge_update, personalization | paper · code · data | 10 conv. 1,305 Q | official_script |
| RoboMemArena | 05 | embodied, multimodal, episodic | paper · code · data | — | official_script |
| SeqMem-Eval | 05 | procedural_workflow, embodied | paper · code | 3,726 Q | official_script |
| SkillEvolBench | 05 | procedural_workflow | paper · code · data | 180 Q | official_script |
| STALE | 05 | semantic_knowledge_update, conversational_longterm, personalization | paper · code · data | 400 conv. 1,200 Q | official_script |
| STATE-Bench | 05 | procedural_workflow | code | 300 conv. 150 Q | official_script |
| SuperMemory-VQA | 05 | episodic, multimodal | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 05 | conversational_longterm, episodic, semantic_knowledge_update, multimodal | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 04 | personalization, semantic_knowledge_update, conversational_longterm | paper | 389 Q | paper_only |
| SEA-Eval | 04 | procedural_workflow | paper · code | 92 conv. 32 Q | data_public |
| SkillFlow | 04 | procedural_workflow | paper · code · data | 166 Q | official_script |
| SkillLearnBench | 04 | procedural_workflow | paper · code | 100 Q | official_script |
| StratMem-Bench | 04 | conversational_longterm | paper · code | 657 Q | data_public |
| AMB (Agent Memory Benchmark) | 03 | conversational_longterm, personalization, episodic, semantic_knowledge_update, coding_cross_session | code · data | — | official_script |
| AMemGym | 03 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 03 | personalization, episodic, multimodal, semantic_knowledge_update | paper · code · data | 1,038 Q | official_script |
| LifeBench | 03 | episodic, personalization, semantic_knowledge_update | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 03 | personalization, conversational_longterm, multimodal, semantic_knowledge_update | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 03 | personalization, conversational_longterm | paper · code | 100 conv. 1,300 Q | official_script |
| RMBench | 03 | embodied, multimodal | paper · code · data | — | official_script |
| AMA-Bench | 02 | episodic, semantic_knowledge_update | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 02 | conversational_longterm, personalization, episodic, semantic_knowledge_update | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 02 | conversational_longterm, semantic_knowledge_update | paper | — | paper_only |
| LoCoMo-Plus | 02 | conversational_longterm, personalization | paper · code | 10 conv. 401 Q | official_script |
| MemoryArena | 02 | episodic, procedural_workflow, personalization | paper · code · data | 701 conv. 4,850 Q | official_script |
| Neuromem | 02 | conversational_longterm, semantic_knowledge_update | paper · code | 10 conv. | official_script |
| PersistBench | 02 | personalization | paper · code | 500 Q | official_script |
| SkillCraft | 02 | procedural_workflow | paper · code | 126 Q | official_script |
| SWE-ContextBench | 02 | coding_cross_session, procedural_workflow | paper · code · data | 376 Q | official_script |
| Trust-Memevo | 02 | procedural_workflow | paper · code | 2,329 Q | paper_only |
| AndroidIntent | 01 | personalization | paper · code · data | 91 conv. 990 Q | official_script |
| CAME-Bench | 01 | conversational_longterm, semantic_knowledge_update | paper · code · data | 14 conv. 373 Q | official_script |
| CloneMem | 01 | personalization, episodic, semantic_knowledge_update | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 01 | conversational_longterm, semantic_knowledge_update, personalization | paper · code | 1,600 conv. | paper_only |
| KnowMe-Bench | 01 | personalization, episodic | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem-Gallery | 01 | conversational_longterm, multimodal, episodic, semantic_knowledge_update | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 01 | conversational_longterm, personalization, semantic_knowledge_update | paper · code | 2,029 conv. 400 Q | data_public |
| MemoryRewardBench | 01 | conversational_longterm | paper · code · data | 2,400 Q | official_script |
| RealMem | 01 | conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code | 10 conv. 1,415 Q | official_script |
| RPEval | 01 | personalization | paper · code | 953 Q | official_script |
| TraineeBench | 01 | procedural_workflow | paper · code | 50 conv. 203 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| PersonaMem-v2 | 12 | personalization, conversational_longterm, memory_security | paper · code · data | 195 conv. 5,000 Q | official_script |
| BELA | 11 | personalization, procedural_workflow | paper · code · data | 40 conv. | official_script |
| ConvoMem | 11 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 75,336 Q | official_script |
| Evo-Memory | 11 | procedural_workflow, embodied | paper | 3,716 Q | paper_only |
| HaluMem | 11 | conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 10 | conversational_longterm, semantic_knowledge_update, episodic, personalization | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemoryBench | 10 | procedural_workflow, conversational_longterm | paper · code · data | 4,063 Q | official_script |
| MemTrack | 10 | semantic_knowledge_update, episodic | paper · data | 47 conv. | data_public |
| Stateful SWE-bench | 10 | coding_cross_session, personalization | paper · code · data | 15 conv. 500 Q | official_script |
| TeleEgo | 10 | episodic, multimodal | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 09 | episodic, multimodal | paper · code | 34 Q | official_script |
| ZH-4O | 09 | conversational_longterm, personalization | paper · code | 28 conv. 1,068 Q | official_script |
| M3-Bench | 08 | episodic, multimodal | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| MPR | 08 | personalization, parametric | paper · code · data | 12 conv. 108,000 Q | official_script |
| OdysseyBench | 08 | conversational_longterm | paper · code | 602 conv. 602 Q | official_script |
| StuLife | 08 | episodic, procedural_workflow | paper · code · data | 1,284 Q | official_script |
| LOCCO | 07 | conversational_longterm, parametric | paper · code | 100 conv. 2,981 Q | data_public |
| MemoryAgentBench | 07 | conversational_longterm, episodic, semantic_knowledge_update | paper · code · data | 2,071 Q | official_script |
| EvaLearn | 06 | procedural_workflow | paper · code | 182 conv. 648 Q | official_script |
| MemBench | 06 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 820 conv. 820 Q | data_public |
| OpenUnlearning | 06 | parametric, memory_security | paper · code · data | 400 Q | official_script |
| StoryBench | 06 | episodic | paper | 86 Q | paper_only |
| SWE-Bench-CL | 06 | coding_cross_session, procedural_workflow | paper · code · data | 8 conv. 273 Q | official_script |
| LifelongAgentBench | 05 | procedural_workflow | paper · code · data | 1,396 Q | official_script |
| MEMENTO | 05 | personalization, episodic, embodied | paper · code | 438 Q | official_script |
| MS-TOD | 05 | conversational_longterm | paper · code | 132 conv. 956 Q | official_script |
| UniEdit | 05 | parametric | paper · code · data | 311,142 Q | official_script |
| WikiDYK | 05 | parametric | paper · code · data | 77,180 Q | official_script |
| KUP (Knowledge Update Playground) | 04 | parametric | paper · code · data | 6,260 Q | official_script |
| PersonaMem | 04 | personalization, conversational_longterm, semantic_knowledge_update | paper · code · data | 60 conv. 2,727 Q | official_script |
| TiC-LM | 04 | parametric | paper · code | — | official_script |
| ImplexConv | 03 | conversational_longterm, personalization | paper · code · data | 2,500 conv. | data_public |
| LifeState-Bench | 03 | episodic, parametric, multi_agent_group, semantic_knowledge_update | paper | 311 Q | paper_only |
| WikiBigEdit | 03 | parametric | paper · code · data | 502,382 Q | official_script |
| MemoryCode | 02 | coding_cross_session, conversational_longterm, semantic_knowledge_update | paper · code · data | 360 conv. | official_script |
| Minerva | 02 | semantic_knowledge_update | paper · code | 1,110 Q | official_script |
| PersonaBench | 02 | personalization, conversational_longterm, semantic_knowledge_update | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 02 | personalization, conversational_longterm | paper · code · data | 3,000 Q | official_script |
| QAEdit | 02 | parametric | paper · code · data | 19,249 Q | official_script |
| REALTALK | 02 | conversational_longterm, episodic | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 02 | conversational_longterm, episodic | paper | 600 Q | paper_only |
| Episodic Memory Benchmark | 01 | episodic, parametric | paper · code · data | 686 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| EvoWiki | 12 | semantic_knowledge_update, parametric | paper · code · data | 10,264 Q | paper_only |
| HalluEditBench | 10 | parametric | paper · code · data | 6,738 Q | official_script |
| LongMemEval | 10 | conversational_longterm, personalization, semantic_knowledge_update, episodic | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 09 | conversational_longterm, semantic_knowledge_update, episodic | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 09 | conversational_longterm, episodic | paper · code | 160 conv. | official_script |
| MemDaily | 09 | personalization | paper · code | 2,954 conv. 2,954 Q | official_script |
| CodeUpdateArena | 07 | parametric | paper · code · data | 670 Q | official_script |
| DialSim | 06 | conversational_longterm, multi_agent_group, episodic | paper · code · data | 3 conv. | official_script |
| StreamBench | 06 | procedural_workflow | paper · code · data | 9,702 Q | official_script |
| TeaFarm | 06 | conversational_longterm | paper · code · data | 100 conv. 200 Q | data_public |
| LoCoMo | 02 | conversational_longterm, episodic | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 02 | personalization, episodic, conversational_longterm | paper · code | 3,409 conv. 8,593 Q | data_public |
| KnowEdit | 01 | parametric | paper · code · data | 5,653 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| AToKe | 12 | parametric | paper · code | 8,820 Q | data_public |
| ConflictEdit / RoundEdit | 10 | parametric | paper · code | 2,500 Q | official_script |
| DMR (Deep Memory Retrieval) | 10 | conversational_longterm | paper · code · data | 500 conv. 500 Q | data_public |
| MMEdit | 10 | parametric, multimodal | paper · code · data | 3,093 Q | official_script |
| RippleEdits | 07 | parametric | paper · code | 4,000 Q | official_script |
| MemoryBank | 05 | conversational_longterm, episodic | paper · code | 15 conv. 194 Q | data_public |
| MQuAKE | 05 | parametric, semantic_knowledge_update | paper · code | 3,000 Q | official_script |
| MemoEval (SCM Self-Controlled Memory evaluation dataset) | 04 | conversational_longterm | paper · code | 18 conv. 105 Q | data_public |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| Multi-Session Chat (MSC) | 07 | conversational_longterm, personalization | paper · code · data | 501 conv. | official_script |
| Level | Meaning |
|---|---|
unverified | No one has opened the paper and repository to check yet |
paper_only | Paper checked; no public data or code found |
data_public | Data is publicly available, but the official repository has no evaluation script |
official_script | The official repository has a script that runs the evaluation, and the data is available |
smoke_tested | We ran the official script on a small sample |
full_run | We ran the full evaluation |
benchmarks/<slug>.yaml following docs/fields.md, and cite a source in sources for every non-null field.python scripts/validate.py and python scripts/build_readme.py.benchmarks/, docs/, schema/ and the READMEs): CC BY 4.0scripts/, templates/, site/, pending/tools/): MITPapers, code and datasets linked from the entries are third-party works under their own licenses, which this repository's licenses do not cover. Check the original source before use.
Structured registry of agent memory benchmarks, with every field traced to its source
HTML
173
25 commits
updated Oct 12, 2026
English | 中文
Browse the registry online → Search and filter by memory type, reproducibility level and year.
A structured registry of agent memory benchmarks: one YAML file per benchmark (benchmarks/<slug>.yaml). This README is generated by a script.
Every field must be traceable to the paper or the official repository; where each value came from is recorded in the entry's sources. Anything that cannot be found is null.
207 entries. Reproducibility levels: unverified 0 · paper_only 37 · data_public 31 · official_script 139 · smoke_tested 0 · full_run 0
A benchmark can belong to several types. Expand a type to see its entries, newest first.
| Type | What the benchmark tests |
|---|---|
conversational_longterm | Recall or reasoning over long, multi-session dialogue or chat histories (e.g. LoCoMo, LongMemEval, MSC) |
personalization | Using user-specific preferences, personas or profiles to tailor answers (e.g. PersonaMem, PrefEval) |
episodic | Recalling specific events and experiences with time, place or order ("what happened when") |
semantic_knowledge_update | Factual knowledge stored outside the weights (context or external store), updated correctly when facts change |
procedural_workflow | Distilling skills, workflows or tool use from earlier tasks and reusing them on different later tasks |
coding_cross_session | Coding agents that must retain information across independent sessions or tasks |
parametric | Memory in the model weights: knowledge editing, injection by fine-tuning, test-time training, memorization or unlearning |
multi_agent_group | Memory shared among several agents, or multi-party / group conversations |
multimodal | Remembered content includes non-text input (images, video, audio) that the system must process directly |
embodied | Agents act in a simulated or real environment; memory is scored through actions or task success |
memory_security | Attacks on or protection of memory: poisoning, injection, privacy leakage, whether deletion requests take effect |
Single-input long-context tests (e.g. RULER, NIAH, LongBench) are not memory benchmarks and are not included.
conversational_longterm (81)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| LoCoMo-Conv | 2026‑09 | paper · code · data | 10 conv. 6,581 Q | official_script |
| LSREP | 2026‑09 | paper · code | 4 conv. 219 Q | paper_only |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| TWIST | 2026‑09 | paper · code · data | 161 Q | official_script |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MemUse | 2026‑08 | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| StateMemBench | 2026‑08 | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 2026‑08 | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| InMind | 2026‑07 | paper · code | 1 conv. 125 Q | data_public |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 2026‑06 | paper · code | 20 conv. 15,422 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| MemAudit | 2026‑05 | paper · code · data | 87 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| RefMem-Bench | 2026‑05 | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| StratMem-Bench | 2026‑04 | paper · code | 657 Q | data_public |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 2026‑03 | paper · code | 100 conv. 1,300 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 2026‑02 | paper | — | paper_only |
| LoCoMo-Plus | 2026‑02 | paper · code | 10 conv. 401 Q | official_script |
| Neuromem | 2026‑02 | paper · code | 10 conv. | official_script |
| CAME-Bench | 2026‑01 | paper · code · data | 14 conv. 373 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| MemoryRewardBench | 2026‑01 | paper · code · data | 2,400 Q | official_script |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemoryBench | 2025‑10 | paper · code · data | 4,063 Q | official_script |
| ZH-4O | 2025‑09 | paper · code | 28 conv. 1,068 Q | official_script |
| OdysseyBench | 2025‑08 | paper · code | 602 conv. 602 Q | official_script |
| LOCCO | 2025‑07 | paper · code | 100 conv. 2,981 Q | data_public |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| MS-TOD | 2025‑05 | paper · code | 132 conv. 956 Q | official_script |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| ImplexConv | 2025‑03 | paper · code · data | 2,500 conv. | data_public |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 2025‑02 | paper · code · data | 3,000 Q | official_script |
| REALTALK | 2025‑02 | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 2025‑02 | paper | 600 Q | paper_only |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 2024‑09 | paper · code | 160 conv. | official_script |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
| TeaFarm | 2024‑06 | paper · code · data | 100 conv. 200 Q | data_public |
| LoCoMo | 2024‑02 | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| DMR (Deep Memory Retrieval) | 2023‑10 | paper · code · data | 500 conv. 500 Q | data_public |
| MemoryBank | 2023‑05 | paper · code | 15 conv. 194 Q | data_public |
| MemoEval (SCM Self-Controlled Memory evaluation dataset) | 2023‑04 | paper · code | 18 conv. 105 Q | data_public |
| Multi-Session Chat (MSC) | 2021‑07 | paper · code · data | 501 conv. | official_script |
personalization (68)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| CMSC-E | 2026‑10 | paper · code · data | — | official_script |
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| Lapse | 2026‑09 | paper | 128 Q | paper_only |
| MemCalib | 2026‑09 | paper · code · data | 1,500 Q | data_public |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| PairPref | 2026‑09 | paper | 1,227 Q | paper_only |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| MemUse | 2026‑08 | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 2026‑08 | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| CAPA | 2026‑07 | paper · code | 60 conv. 300 Q | data_public |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| InMind | 2026‑07 | paper · code | 1 conv. 125 Q | data_public |
| MemSyco-Bench | 2026‑07 | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 2026‑07 | paper · code | — | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MIST | 2026‑06 | paper | 598 conv. 598 Q | paper_only |
| StreamMemBench | 2026‑06 | paper · code | 16,214 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| RefMem-Bench | 2026‑05 | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 2026‑03 | paper · code | 100 conv. 1,300 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| LoCoMo-Plus | 2026‑02 | paper · code | 10 conv. 401 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| PersistBench | 2026‑02 | paper · code | 500 Q | official_script |
| AndroidIntent | 2026‑01 | paper · code · data | 91 conv. 990 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| KnowMe-Bench | 2026‑01 | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| RPEval | 2026‑01 | paper · code | 953 Q | official_script |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| BELA | 2025‑11 | paper · code · data | 40 conv. | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| Stateful SWE-bench | 2025‑10 | paper · code · data | 15 conv. 500 Q | official_script |
| ZH-4O | 2025‑09 | paper · code | 28 conv. 1,068 Q | official_script |
| MPR | 2025‑08 | paper · code · data | 12 conv. 108,000 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| ImplexConv | 2025‑03 | paper · code · data | 2,500 conv. | data_public |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 2025‑02 | paper · code · data | 3,000 Q | official_script |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| MemDaily | 2024‑09 | paper · code | 2,954 conv. 2,954 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| Multi-Session Chat (MSC) | 2021‑07 | paper · code · data | 501 conv. | official_script |
episodic (56)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| RealCompanion | 2026‑10 | paper · code · data | 10 conv. 3,235 Q | paper_only |
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| DolphinBench | 2026‑09 | paper · code · data | 3 conv. 600 Q | official_script |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| ICM-Bench | 2026‑09 | paper · code · data | 839 conv. 1,217 Q | official_script |
| MemTransfer | 2026‑09 | paper | 100 conv. 100 Q | paper_only |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| AgenticSTS | 2026‑07 | paper · code · data | — | official_script |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| MERIT | 2026‑07 | paper · code | — | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| DMV-Bench | 2026‑06 | paper · code · data | 35,410 Q | official_script |
| MemAudit (formerly MEMPROBE) | 2026‑06 | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| StreamMemBench | 2026‑06 | paper · code | 16,214 Q | official_script |
| EgoStream | 2026‑05 | paper | 2,250 Q | paper_only |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SuperMemory-VQA | 2026‑05 | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| AMA-Bench | 2026‑02 | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| KnowMe-Bench | 2026‑01 | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemTrack | 2025‑10 | paper · data | 47 conv. | data_public |
| TeleEgo | 2025‑10 | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 2025‑09 | paper · code | 34 Q | official_script |
| M3-Bench | 2025‑08 | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| StuLife | 2025‑08 | paper · code · data | 1,284 Q | official_script |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| StoryBench | 2025‑06 | paper | 86 Q | paper_only |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| REALTALK | 2025‑02 | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 2025‑02 | paper | 600 Q | paper_only |
| Episodic Memory Benchmark | 2025‑01 | paper · code · data | 686 Q | official_script |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 2024‑09 | paper · code | 160 conv. | official_script |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
| LoCoMo | 2024‑02 | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 2024‑02 | paper · code | 3,409 conv. 8,593 Q | data_public |
| MemoryBank | 2023‑05 | paper · code | 15 conv. 194 Q | data_public |
semantic_knowledge_update (74)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| DyadMem | 2026‑10 | paper | 3,065 conv. 61,210 Q | paper_only |
| Knowing the Store | 2026‑10 | paper · code · data | 444 Q | official_script |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| Controlled In-Context Memory | 2026‑09 | paper | 1,200 conv. 1,200 Q | paper_only |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| EvoPathBench | 2026‑09 | paper · code | 1,800 conv. 3,600 Q | official_script |
| Impact Is Not Invalidation | 2026‑09 | paper | 10,369 Q | paper_only |
| Lapse | 2026‑09 | paper | 128 Q | paper_only |
| LSREP | 2026‑09 | paper · code | 4 conv. 219 Q | paper_only |
| MemProbe | 2026‑09 | paper · code · data | 56 conv. 224 Q | official_script |
| MemRiskBench | 2026‑09 | paper · code · data | 120 conv. 120 Q | official_script |
| PRAGMA | 2026‑09 | paper · code · data | 100 conv. 400 Q | data_public |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| agent-memory-bench | 2026‑08 | code · data | 196 conv. 26 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| StateMemBench | 2026‑08 | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 2026‑08 | paper | 40 conv. 541 Q | paper_only |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MemSyco-Bench | 2026‑07 | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 2026‑07 | paper · code | — | official_script |
| PM-Bench | 2026‑07 | paper · code | 1 conv. 81 Q | official_script |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| ForgetEval | 2026‑06 | paper · code | 385 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 2026‑06 | paper · code | 20 conv. 15,422 Q | official_script |
| SubtleMemory | 2026‑06 | paper · code · data | 10 conv. 1,522 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| MemAudit | 2026‑05 | paper · code · data | 87 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 2026‑05 | paper · code · data | 161 Q | official_script |
| PrecisionMemBench | 2026‑05 | paper · code · data | 2 conv. 89 Q | official_script |
| RHELM | 2026‑05 | paper · code · data | 10 conv. 1,305 Q | official_script |
| STALE | 2026‑05 | paper · code · data | 400 conv. 1,200 Q | official_script |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 2026‑04 | paper | 389 Q | paper_only |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| AMemGym | 2026‑03 | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| LifeBench | 2026‑03 | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| AMA-Bench | 2026‑02 | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 2026‑02 | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 2026‑02 | paper | — | paper_only |
| Neuromem | 2026‑02 | paper · code | 10 conv. | official_script |
| CAME-Bench | 2026‑01 | paper · code · data | 14 conv. 373 Q | official_script |
| CloneMem | 2026‑01 | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 2026‑01 | paper · code | 1,600 conv. | paper_only |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 2026‑01 | paper · code | 2,029 conv. 400 Q | data_public |
| RealMem | 2026‑01 | paper · code | 10 conv. 1,415 Q | official_script |
| ConvoMem | 2025‑11 | paper · code · data | 75,336 Q | official_script |
| HaluMem | 2025‑11 | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 2025‑10 | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemTrack | 2025‑10 | paper · data | 47 conv. | data_public |
| MemoryAgentBench | 2025‑07 | paper · code · data | 2,071 Q | official_script |
| MemBench | 2025‑06 | paper · code · data | 820 conv. 820 Q | data_public |
| PersonaMem | 2025‑04 | paper · code · data | 60 conv. 2,727 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
| Minerva | 2025‑02 | paper · code | 1,110 Q | official_script |
| PersonaBench | 2025‑02 | paper · code | 15 conv. 582 Q | official_script |
| EvoWiki | 2024‑12 | paper · code · data | 10,264 Q | paper_only |
| LongMemEval | 2024‑10 | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 2024‑09 | paper · code | 1 conv. 33 Q | official_script |
| MQuAKE | 2023‑05 | paper · code | 3,000 Q | official_script |
procedural_workflow (53)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| Learn2Play Bench | 2026‑10 | paper · code | 20 Q | data_public |
| MetaKernelBench | 2026‑10 | paper · code | 74 Q | official_script |
| MLE-InsightBench | 2026‑10 | paper | 12 Q | paper_only |
| ServeLearnBench | 2026‑10 | paper · code | 3,210 Q | official_script |
| ConsolidationBench | 2026‑09 | paper | — | paper_only |
| EconSkills | 2026‑09 | paper · data | 360 Q | data_public |
| EngramBench | 2026‑09 | paper · code | 30 conv. 13 Q | official_script |
| EvoHarnessBench | 2026‑09 | paper · data | 802 Q | data_public |
| EvoPathBench | 2026‑09 | paper · code | 1,800 conv. 3,600 Q | official_script |
| VibeMemBench | 2026‑09 | paper · code | 3,634 conv. 111 Q | paper_only |
| ContinualSkillBench | 2026‑08 | paper · code | 500 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| FinEvo-Bench | 2026‑08 | paper | 120 Q | paper_only |
| GDPevo | 2026‑08 | paper · code | 240 Q | official_script |
| MemTrapBench | 2026‑08 | paper · code | 1,050 conv. 1,050 Q | official_script |
| PAST-Bench | 2026‑08 | paper · code | 26 conv. 204 Q | official_script |
| PATH-Bench | 2026‑08 | paper | 120 Q | paper_only |
| S3Gym | 2026‑08 | paper | — | paper_only |
| AgenticSTS | 2026‑07 | paper · code · data | — | official_script |
| AgentStream | 2026‑07 | paper · code | 300 Q | official_script |
| EvoAgentBench | 2026‑07 | paper · code · data | 267 Q | official_script |
| SLVMBench | 2026‑07 | paper · data | 2,261 Q | data_public |
| AFTER | 2026‑06 | paper · code · data | 382 Q | data_public |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| AhaBench | 2026‑06 | paper · code | — | official_script |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| FinEvolveBench | 2026‑06 | paper · code · data | — | official_script |
| MedEvoEval | 2026‑06 | paper | 700 Q | paper_only |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| SEAGym | 2026‑06 | paper · code | 55 Q | official_script |
| BenchTrace | 2026‑05 | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| SeqMem-Eval | 2026‑05 | paper · code | 3,726 Q | official_script |
| SkillEvolBench | 2026‑05 | paper · code · data | 180 Q | official_script |
| STATE-Bench | 2026‑05 | code | 300 conv. 150 Q | official_script |
| SEA-Eval | 2026‑04 | paper · code | 92 conv. 32 Q | data_public |
| SkillFlow | 2026‑04 | paper · code · data | 166 Q | official_script |
| SkillLearnBench | 2026‑04 | paper · code | 100 Q | official_script |
| MemoryArena | 2026‑02 | paper · code · data | 701 conv. 4,850 Q | official_script |
| SkillCraft | 2026‑02 | paper · code | 126 Q | official_script |
| SWE-ContextBench | 2026‑02 | paper · code · data | 376 Q | official_script |
| Trust-Memevo | 2026‑02 | paper · code | 2,329 Q | paper_only |
| TraineeBench | 2026‑01 | paper · code | 50 conv. 203 Q | official_script |
| BELA | 2025‑11 | paper · code · data | 40 conv. | official_script |
| Evo-Memory | 2025‑11 | paper | 3,716 Q | paper_only |
| MemoryBench | 2025‑10 | paper · code · data | 4,063 Q | official_script |
| StuLife | 2025‑08 | paper · code · data | 1,284 Q | official_script |
| EvaLearn | 2025‑06 | paper · code | 182 conv. 648 Q | official_script |
| SWE-Bench-CL | 2025‑06 | paper · code · data | 8 conv. 273 Q | official_script |
| LifelongAgentBench | 2025‑05 | paper · code · data | 1,396 Q | official_script |
| StreamBench | 2024‑06 | paper · code · data | 9,702 Q | official_script |
coding_cross_session (17)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| MetaKernelBench | 2026‑10 | paper · code | 74 Q | official_script |
| EngramBench | 2026‑09 | paper · code | 30 conv. 13 Q | official_script |
| VibeMemBench | 2026‑09 | paper · code | 3,634 conv. 111 Q | paper_only |
| agent-memory-bench | 2026‑08 | code · data | 196 conv. 26 Q | official_script |
| DreamBench-SWE | 2026‑08 | paper · code | 60 conv. 60 Q | data_public |
| CAPA | 2026‑07 | paper · code | 60 conv. 300 Q | data_public |
| ChainSWE | 2026‑07 | paper · data | 100 conv. 304 Q | data_public |
| EvoAgentBench | 2026‑07 | paper · code · data | 267 Q | official_script |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 2026‑07 | paper · code | 194 Q | official_script |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| Continual Learning Bench (CL-Bench) | 2026‑06 | paper · code · data | 6 conv. 301 Q | official_script |
| EvoArena | 2026‑06 | paper · code · data | 10 conv. 505 Q | official_script |
| AMB (Agent Memory Benchmark) | 2026‑03 | code · data | — | official_script |
| SWE-ContextBench | 2026‑02 | paper · code · data | 376 Q | official_script |
| Stateful SWE-bench | 2025‑10 | paper · code · data | 15 conv. 500 Q | official_script |
| SWE-Bench-CL | 2025‑06 | paper · code · data | 8 conv. 273 Q | official_script |
| MemoryCode | 2025‑02 | paper · code · data | 360 conv. | official_script |
parametric (27)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| Cross-lingual Unlearning Tensor | 2026‑09 | paper · code | 4,000 Q | paper_only |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| MemoReason | 2026‑09 | paper · code · data | 1,200 Q | official_script |
| MemToC | 2026‑08 | paper · code | 6,504 Q | data_public |
| S3Gym | 2026‑08 | paper | — | paper_only |
| ForgetBench | 2026‑07 | paper | 6,431 Q | paper_only |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MPR | 2025‑08 | paper · code · data | 12 conv. 108,000 Q | official_script |
| LOCCO | 2025‑07 | paper · code | 100 conv. 2,981 Q | data_public |
| OpenUnlearning | 2025‑06 | paper · code · data | 400 Q | official_script |
| UniEdit | 2025‑05 | paper · code · data | 311,142 Q | official_script |
| WikiDYK | 2025‑05 | paper · code · data | 77,180 Q | official_script |
| KUP (Knowledge Update Playground) | 2025‑04 | paper · code · data | 6,260 Q | official_script |
| TiC-LM | 2025‑04 | paper · code | — | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| WikiBigEdit | 2025‑03 | paper · code · data | 502,382 Q | official_script |
| QAEdit | 2025‑02 | paper · code · data | 19,249 Q | official_script |
| Episodic Memory Benchmark | 2025‑01 | paper · code · data | 686 Q | official_script |
| EvoWiki | 2024‑12 | paper · code · data | 10,264 Q | paper_only |
| HalluEditBench | 2024‑10 | paper · code · data | 6,738 Q | official_script |
| CodeUpdateArena | 2024‑07 | paper · code · data | 670 Q | official_script |
| KnowEdit | 2024‑01 | paper · code · data | 5,653 Q | official_script |
| AToKe | 2023‑12 | paper · code | 8,820 Q | data_public |
| ConflictEdit / RoundEdit | 2023‑10 | paper · code | 2,500 Q | official_script |
| MMEdit | 2023‑10 | paper · code · data | 3,093 Q | official_script |
| RippleEdits | 2023‑07 | paper · code | 4,000 Q | official_script |
| MQuAKE | 2023‑05 | paper · code | 3,000 Q | official_script |
multi_agent_group (8)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| CoMemBench | 2026‑09 | paper | 800 Q | paper_only |
| Correlated Promotion Benchmark (CPB) | 2026‑09 | paper · code · data | 100 conv. | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| MemoryDocDataSet | 2026‑06 | paper | 50 conv. 1,000 Q | paper_only |
| GroupMemBench | 2026‑05 | paper · code · data | 4 conv. 745 Q | official_script |
| LifeState-Bench | 2025‑03 | paper | 311 Q | paper_only |
| DialSim | 2024‑06 | paper · code · data | 3 conv. | official_script |
multimodal (28)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| APM-Bench | 2026‑09 | paper · code · data | 104 conv. 2,719 Q | paper_only |
| CUE-Mem | 2026‑09 | paper · code · data | 648 conv. 2,674 Q | official_script |
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| ICM-Bench | 2026‑09 | paper · code · data | 839 conv. 1,217 Q | official_script |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| Video2Skill | 2026‑09 | paper · data | 172 conv. 1,671 Q | data_public |
| VoxMem | 2026‑09 | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| MobileMem | 2026‑08 | paper · code · data | 16 conv. 7,415 Q | official_script |
| Sequential-EQA | 2026‑07 | paper · code | 50 conv. 498 Q | official_script |
| SLVMBench | 2026‑07 | paper · data | 2,261 Q | data_public |
| DMV-Bench | 2026‑06 | paper · code · data | 35,410 Q | official_script |
| H2HMem | 2026‑06 | paper · code · data | 25 conv. 2,236 Q | official_script |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| EgoStream | 2026‑05 | paper | 2,250 Q | paper_only |
| LongMemEval-V2 | 2026‑05 | paper · code · data | 1,540 conv. 451 Q | official_script |
| MBench | 2026‑05 | paper · code · data | 1,040 Q | official_script |
| MemEye | 2026‑05 | paper · code · data | 8 conv. 371 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SuperMemory-VQA | 2026‑05 | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 2026‑05 | paper · code · data | 461 conv. 24,258 Q | official_script |
| ATM-Bench | 2026‑03 | paper · code · data | 1,038 Q | official_script |
| Persona-MME | 2026‑03 | paper · code · data | 200 conv. 2,034 Q | official_script |
| RMBench | 2026‑03 | paper · code · data | — | official_script |
| Mem-Gallery | 2026‑01 | paper · code · data | 20 conv. 1,711 Q | official_script |
| TeleEgo | 2025‑10 | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 2025‑09 | paper · code | 34 Q | official_script |
| M3-Bench | 2025‑08 | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| MMEdit | 2023‑10 | paper · code · data | 3,093 Q | official_script |
embodied (12)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| EmbodiedMemory-Bench (EMem-Bench) | 2026‑09 | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| MemTransfer | 2026‑09 | paper | 100 conv. 100 Q | paper_only |
| Sequential-EQA | 2026‑07 | paper · code | 50 conv. 498 Q | official_script |
| AgentCL | 2026‑06 | paper · data | — | data_public |
| OmniGameArena | 2026‑06 | paper · code · data | — | official_script |
| BenchTrace | 2026‑05 | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EvoMemBench | 2026‑05 | paper · code | 5,754 Q | official_script |
| RoboMemArena | 2026‑05 | paper · code · data | — | official_script |
| SeqMem-Eval | 2026‑05 | paper · code | 3,726 Q | official_script |
| RMBench | 2026‑03 | paper · code · data | — | official_script |
| Evo-Memory | 2025‑11 | paper | 3,716 Q | paper_only |
| MEMENTO | 2025‑05 | paper · code | 438 Q | official_script |
memory_security (18)| Name | Date | Links | Scale | Repro level |
|---|---|---|---|---|
| ConsolidationBench | 2026‑09 | paper | — | paper_only |
| Correlated Promotion Benchmark (CPB) | 2026‑09 | paper · code · data | 100 conv. | official_script |
| Cross-lingual Unlearning Tensor | 2026‑09 | paper · code | 4,000 Q | paper_only |
| IDUnlearn-Bench | 2026‑09 | paper · code · data | 4,860 Q | data_public |
| MemRiskBench | 2026‑09 | paper · code · data | 120 conv. 120 Q | official_script |
| TWIST | 2026‑09 | paper · code · data | 161 Q | official_script |
| MemTrapBench | 2026‑08 | paper · code | 1,050 conv. 1,050 Q | official_script |
| Ground Truth First | 2026‑07 | paper · code | 383 Q | paper_only |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 2026‑07 | paper · code | 194 Q | official_script |
| MemOps | 2026‑07 | paper · code | 403 conv. 2,006 Q | official_script |
| MemSecBench | 2026‑07 | paper | 310 Q | paper_only |
| RUMBA | 2026‑07 | paper · code · data | 85 conv. 1,543 Q | official_script |
| ForgetEval | 2026‑06 | paper · code | 385 Q | official_script |
| GateMem | 2026‑06 | paper · code · data | 91 conv. 2,218 Q | official_script |
| HUSH-Bench | 2026‑06 | paper | 10 conv. 2,400 Q | paper_only |
| MPBench | 2026‑06 | paper · code | 3,240 Q | data_public |
| PersonaMem-v2 | 2025‑12 | paper · code · data | 195 conv. 5,000 Q | official_script |
| OpenUnlearning | 2025‑06 | paper · code · data | 400 Q | official_script |
Grouped by year, newest first. Click a name to see all fields and their sources.
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| CMSC-E | 10 | personalization | paper · code · data | — | official_script |
| DyadMem | 10 | conversational_longterm, personalization, episodic, semantic_knowledge_update | paper | 3,065 conv. 61,210 Q | paper_only |
| Knowing the Store | 10 | semantic_knowledge_update | paper · code · data | 444 Q | official_script |
| Learn2Play Bench | 10 | procedural_workflow | paper · code | 20 Q | data_public |
| MetaKernelBench | 10 | procedural_workflow, coding_cross_session | paper · code | 74 Q | official_script |
| MLE-InsightBench | 10 | procedural_workflow | paper | 12 Q | paper_only |
| RealCompanion | 10 | conversational_longterm, personalization, episodic | paper · code · data | 10 conv. 3,235 Q | paper_only |
| ServeLearnBench | 10 | procedural_workflow, semantic_knowledge_update, personalization | paper · code | 3,210 Q | official_script |
| APM-Bench | 09 | episodic, semantic_knowledge_update, multimodal | paper · code · data | 104 conv. 2,719 Q | paper_only |
| CoMemBench | 09 | multi_agent_group | paper | 800 Q | paper_only |
| ConsolidationBench | 09 | procedural_workflow, memory_security | paper | — | paper_only |
| Controlled In-Context Memory | 09 | semantic_knowledge_update | paper | 1,200 conv. 1,200 Q | paper_only |
| Correlated Promotion Benchmark (CPB) | 09 | multi_agent_group, memory_security | paper · code · data | 100 conv. | official_script |
| Cross-lingual Unlearning Tensor | 09 | parametric, memory_security | paper · code | 4,000 Q | paper_only |
| CUE-Mem | 09 | conversational_longterm, personalization, multimodal | paper · code · data | 648 conv. 2,674 Q | official_script |
| DolphinBench | 09 | conversational_longterm, personalization, episodic | paper · code · data | 3 conv. 600 Q | official_script |
| EconSkills | 09 | procedural_workflow | paper · data | 360 Q | data_public |
| EmbodiedMemory-Bench (EMem-Bench) | 09 | embodied, episodic, semantic_knowledge_update, multimodal | paper · code · data | 2,554 conv. 2,554 Q | official_script |
| EngramBench | 09 | procedural_workflow, coding_cross_session | paper · code | 30 conv. 13 Q | official_script |
| EvoHarnessBench | 09 | procedural_workflow | paper · data | 802 Q | data_public |
| EvoPathBench | 09 | procedural_workflow, semantic_knowledge_update | paper · code | 1,800 conv. 3,600 Q | official_script |
| ICM-Bench | 09 | episodic, multimodal | paper · code · data | 839 conv. 1,217 Q | official_script |
| IDUnlearn-Bench | 09 | parametric, memory_security, multimodal | paper · code · data | 4,860 Q | data_public |
| Impact Is Not Invalidation | 09 | semantic_knowledge_update | paper | 10,369 Q | paper_only |
| Lapse | 09 | personalization, semantic_knowledge_update | paper | 128 Q | paper_only |
| LoCoMo-Conv | 09 | conversational_longterm | paper · code · data | 10 conv. 6,581 Q | official_script |
| LSREP | 09 | conversational_longterm, semantic_knowledge_update | paper · code | 4 conv. 219 Q | paper_only |
| MemCalib | 09 | personalization | paper · code · data | 1,500 Q | data_public |
| MemoReason | 09 | parametric | paper · code · data | 1,200 Q | official_script |
| MemProbe | 09 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 56 conv. 224 Q | official_script |
| MemRiskBench | 09 | semantic_knowledge_update, memory_security | paper · code · data | 120 conv. 120 Q | official_script |
| MemTransfer | 09 | episodic, embodied | paper | 100 conv. 100 Q | paper_only |
| PairPref | 09 | personalization | paper | 1,227 Q | paper_only |
| PRAGMA | 09 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 100 conv. 400 Q | data_public |
| TWIST | 09 | conversational_longterm, memory_security | paper · code · data | 161 Q | official_script |
| VibeMemBench | 09 | coding_cross_session, procedural_workflow | paper · code | 3,634 conv. 111 Q | paper_only |
| Video2Skill | 09 | multimodal | paper · data | 172 conv. 1,671 Q | data_public |
| VoxMem | 09 | conversational_longterm, multimodal, semantic_knowledge_update, episodic | paper · code · data | 3,196 conv. 3,196 Q | official_script |
| agent-memory-bench | 08 | coding_cross_session, semantic_knowledge_update | code · data | 196 conv. 26 Q | official_script |
| ContinualSkillBench | 08 | procedural_workflow | paper · code | 500 Q | official_script |
| DreamBench-SWE | 08 | coding_cross_session, semantic_knowledge_update, procedural_workflow | paper · code | 60 conv. 60 Q | data_public |
| FinEvo-Bench | 08 | procedural_workflow | paper | 120 Q | paper_only |
| GDPevo | 08 | procedural_workflow | paper · code | 240 Q | official_script |
| MemToC | 08 | parametric | paper · code | 6,504 Q | data_public |
| MemTrapBench | 08 | procedural_workflow, memory_security | paper · code | 1,050 conv. 1,050 Q | official_script |
| MemUse | 08 | conversational_longterm, personalization | paper · code · data | 72 conv. 316 Q | official_script |
| MobileMem | 08 | conversational_longterm, personalization, episodic, semantic_knowledge_update, multimodal | paper · code · data | 16 conv. 7,415 Q | official_script |
| PAST-Bench | 08 | procedural_workflow, semantic_knowledge_update, personalization, episodic | paper · code | 26 conv. 204 Q | official_script |
| PATH-Bench | 08 | procedural_workflow | paper | 120 Q | paper_only |
| S3Gym | 08 | procedural_workflow, parametric | paper | — | paper_only |
| StateMemBench | 08 | conversational_longterm, semantic_knowledge_update | paper | 234 conv. 322 Q | paper_only |
| TANGLE | 08 | personalization, conversational_longterm, semantic_knowledge_update | paper | 40 conv. 541 Q | paper_only |
| UTILMEM | 08 | conversational_longterm, personalization | paper · code · data | 1,717 conv. 1,717 Q | data_public |
| AgenticSTS | 07 | procedural_workflow, episodic | paper · code · data | — | official_script |
| AgentStream | 07 | procedural_workflow | paper · code | 300 Q | official_script |
| CAPA | 07 | coding_cross_session, personalization | paper · code | 60 conv. 300 Q | data_public |
| ChainSWE | 07 | coding_cross_session | paper · data | 100 conv. 304 Q | data_public |
| EvoAgentBench | 07 | procedural_workflow, coding_cross_session | paper · code · data | 267 Q | official_script |
| ForgetBench | 07 | parametric | paper | 6,431 Q | paper_only |
| Ground Truth First | 07 | conversational_longterm, personalization, episodic, semantic_knowledge_update, memory_security | paper · code | 383 Q | paper_only |
| InMind | 07 | personalization, conversational_longterm | paper · code | 1 conv. 125 Q | data_public |
| Insecure Coding Preferences in Long-Term Memory (MemSecurity) | 07 | coding_cross_session, memory_security | paper · code | 194 Q | official_script |
| MemOps | 07 | conversational_longterm, semantic_knowledge_update, parametric, memory_security | paper · code | 403 conv. 2,006 Q | official_script |
| MemSecBench | 07 | memory_security | paper | 310 Q | paper_only |
| MemSyco-Bench | 07 | personalization, semantic_knowledge_update | paper · code · data | 1,550 conv. 1,550 Q | official_script |
| MERIT | 07 | episodic, semantic_knowledge_update, personalization | paper · code | — | official_script |
| PM-Bench | 07 | semantic_knowledge_update | paper · code | 1 conv. 81 Q | official_script |
| RUMBA | 07 | conversational_longterm, episodic, semantic_knowledge_update, memory_security | paper · code · data | 85 conv. 1,543 Q | official_script |
| Sequential-EQA | 07 | embodied, multimodal | paper · code | 50 conv. 498 Q | official_script |
| SLVMBench | 07 | procedural_workflow, multimodal | paper · data | 2,261 Q | data_public |
| AFTER | 06 | procedural_workflow | paper · code · data | 382 Q | data_public |
| AgentCL | 06 | procedural_workflow, coding_cross_session, embodied | paper · data | — | data_public |
| AhaBench | 06 | procedural_workflow | paper · code | — | official_script |
| Continual Learning Bench (CL-Bench) | 06 | procedural_workflow, coding_cross_session, semantic_knowledge_update | paper · code · data | 6 conv. 301 Q | official_script |
| DMV-Bench | 06 | episodic, multimodal | paper · code · data | 35,410 Q | official_script |
| EvoArena | 06 | semantic_knowledge_update, procedural_workflow, coding_cross_session, personalization, conversational_longterm | paper · code · data | 10 conv. 505 Q | official_script |
| FinEvolveBench | 06 | procedural_workflow | paper · code · data | — | official_script |
| ForgetEval | 06 | semantic_knowledge_update, memory_security | paper · code | 385 Q | official_script |
| GateMem | 06 | multi_agent_group, memory_security, semantic_knowledge_update, conversational_longterm | paper · code · data | 91 conv. 2,218 Q | official_script |
| H2HMem | 06 | conversational_longterm, multimodal, multi_agent_group, semantic_knowledge_update | paper · code · data | 25 conv. 2,236 Q | official_script |
| HUSH-Bench | 06 | memory_security, personalization, conversational_longterm | paper | 10 conv. 2,400 Q | paper_only |
| MedEvoEval | 06 | procedural_workflow | paper | 700 Q | paper_only |
| MemAudit (formerly MEMPROBE) | 06 | personalization, conversational_longterm, episodic | paper · code | 50 conv. 1,550 Q | official_script |
| MemoryDocDataSet | 06 | conversational_longterm, multi_agent_group, semantic_knowledge_update, episodic | paper | 50 conv. 1,000 Q | paper_only |
| MemTrace | 06 | conversational_longterm, semantic_knowledge_update | paper · code | 20 conv. 15,422 Q | official_script |
| MIST | 06 | personalization | paper | 598 conv. 598 Q | paper_only |
| MPBench | 06 | memory_security | paper · code | 3,240 Q | data_public |
| OmniGameArena | 06 | procedural_workflow, embodied, multimodal | paper · code · data | — | official_script |
| SEAGym | 06 | procedural_workflow | paper · code | 55 Q | official_script |
| StreamMemBench | 06 | personalization, episodic | paper · code | 16,214 Q | official_script |
| SubtleMemory | 06 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 10 conv. 1,522 Q | official_script |
| BenchTrace | 05 | procedural_workflow, embodied | paper · code · data | 1,821 conv. 4,391 Q | official_script |
| EgoStream | 05 | episodic, multimodal | paper | 2,250 Q | paper_only |
| EvoMemBench | 05 | procedural_workflow, semantic_knowledge_update, embodied, conversational_longterm | paper · code | 5,754 Q | official_script |
| GroupMemBench | 05 | multi_agent_group, conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code · data | 4 conv. 745 Q | official_script |
| LongMemEval-V2 | 05 | episodic, procedural_workflow, multimodal | paper · code · data | 1,540 conv. 451 Q | official_script |
| MBench | 05 | multimodal | paper · code · data | 1,040 Q | official_script |
| MemAudit | 05 | semantic_knowledge_update, conversational_longterm | paper · code · data | 87 Q | official_script |
| MemEye | 05 | conversational_longterm, multimodal, semantic_knowledge_update | paper · code · data | 8 conv. 371 Q | official_script |
| Momento | 05 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 161 Q | official_script |
| PrecisionMemBench | 05 | semantic_knowledge_update | paper · code · data | 2 conv. 89 Q | official_script |
| RefMem-Bench | 05 | conversational_longterm, personalization | paper | 35 conv. 20,279 Q | paper_only |
| RHELM | 05 | conversational_longterm, episodic, semantic_knowledge_update, personalization | paper · code · data | 10 conv. 1,305 Q | official_script |
| RoboMemArena | 05 | embodied, multimodal, episodic | paper · code · data | — | official_script |
| SeqMem-Eval | 05 | procedural_workflow, embodied | paper · code | 3,726 Q | official_script |
| SkillEvolBench | 05 | procedural_workflow | paper · code · data | 180 Q | official_script |
| STALE | 05 | semantic_knowledge_update, conversational_longterm, personalization | paper · code · data | 400 conv. 1,200 Q | official_script |
| STATE-Bench | 05 | procedural_workflow | code | 300 conv. 150 Q | official_script |
| SuperMemory-VQA | 05 | episodic, multimodal | paper · code · data | 4,853 Q | data_public |
| WorldMemArena | 05 | conversational_longterm, episodic, semantic_knowledge_update, multimodal | paper · code · data | 461 conv. 24,258 Q | official_script |
| MemHome / MemHomeLife | 04 | personalization, semantic_knowledge_update, conversational_longterm | paper | 389 Q | paper_only |
| SEA-Eval | 04 | procedural_workflow | paper · code | 92 conv. 32 Q | data_public |
| SkillFlow | 04 | procedural_workflow | paper · code · data | 166 Q | official_script |
| SkillLearnBench | 04 | procedural_workflow | paper · code | 100 Q | official_script |
| StratMem-Bench | 04 | conversational_longterm | paper · code | 657 Q | data_public |
| AMB (Agent Memory Benchmark) | 03 | conversational_longterm, personalization, episodic, semantic_knowledge_update, coding_cross_session | code · data | — | official_script |
| AMemGym | 03 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 20 conv. 200 Q | official_script |
| ATM-Bench | 03 | personalization, episodic, multimodal, semantic_knowledge_update | paper · code · data | 1,038 Q | official_script |
| LifeBench | 03 | episodic, personalization, semantic_knowledge_update | paper · code · data | 10 conv. 2,003 Q | official_script |
| Persona-MME | 03 | personalization, conversational_longterm, multimodal, semantic_knowledge_update | paper · code · data | 200 conv. 2,034 Q | official_script |
| RealPref | 03 | personalization, conversational_longterm | paper · code | 100 conv. 1,300 Q | official_script |
| RMBench | 03 | embodied, multimodal | paper · code · data | — | official_script |
| AMA-Bench | 02 | episodic, semantic_knowledge_update | paper · code · data | 208 conv. 2,496 Q | official_script |
| ES-MemEval | 02 | conversational_longterm, personalization, episodic, semantic_knowledge_update | paper · code · data | 18 conv. 1,209 Q | official_script |
| HLME | 02 | conversational_longterm, semantic_knowledge_update | paper | — | paper_only |
| LoCoMo-Plus | 02 | conversational_longterm, personalization | paper · code | 10 conv. 401 Q | official_script |
| MemoryArena | 02 | episodic, procedural_workflow, personalization | paper · code · data | 701 conv. 4,850 Q | official_script |
| Neuromem | 02 | conversational_longterm, semantic_knowledge_update | paper · code | 10 conv. | official_script |
| PersistBench | 02 | personalization | paper · code | 500 Q | official_script |
| SkillCraft | 02 | procedural_workflow | paper · code | 126 Q | official_script |
| SWE-ContextBench | 02 | coding_cross_session, procedural_workflow | paper · code · data | 376 Q | official_script |
| Trust-Memevo | 02 | procedural_workflow | paper · code | 2,329 Q | paper_only |
| AndroidIntent | 01 | personalization | paper · code · data | 91 conv. 990 Q | official_script |
| CAME-Bench | 01 | conversational_longterm, semantic_knowledge_update | paper · code · data | 14 conv. 373 Q | official_script |
| CloneMem | 01 | personalization, episodic, semantic_knowledge_update | paper · code · data | 10 conv. 1,183 Q | official_script |
| EvolMem | 01 | conversational_longterm, semantic_knowledge_update, personalization | paper · code | 1,600 conv. | paper_only |
| KnowMe-Bench | 01 | personalization, episodic | paper · code · data | 3 conv. 2,580 Q | official_script |
| Mem-Gallery | 01 | conversational_longterm, multimodal, episodic, semantic_knowledge_update | paper · code · data | 20 conv. 1,711 Q | official_script |
| Mem2ActBench | 01 | conversational_longterm, personalization, semantic_knowledge_update | paper · code | 2,029 conv. 400 Q | data_public |
| MemoryRewardBench | 01 | conversational_longterm | paper · code · data | 2,400 Q | official_script |
| RealMem | 01 | conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code | 10 conv. 1,415 Q | official_script |
| RPEval | 01 | personalization | paper · code | 953 Q | official_script |
| TraineeBench | 01 | procedural_workflow | paper · code | 50 conv. 203 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| PersonaMem-v2 | 12 | personalization, conversational_longterm, memory_security | paper · code · data | 195 conv. 5,000 Q | official_script |
| BELA | 11 | personalization, procedural_workflow | paper · code · data | 40 conv. | official_script |
| ConvoMem | 11 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 75,336 Q | official_script |
| Evo-Memory | 11 | procedural_workflow, embodied | paper | 3,716 Q | paper_only |
| HaluMem | 11 | conversational_longterm, semantic_knowledge_update, personalization, episodic | paper · code · data | 20 conv. 3,467 Q | official_script |
| BEAM | 10 | conversational_longterm, semantic_knowledge_update, episodic, personalization | paper · code · data | 100 conv. 2,000 Q | official_script |
| MemoryBench | 10 | procedural_workflow, conversational_longterm | paper · code · data | 4,063 Q | official_script |
| MemTrack | 10 | semantic_knowledge_update, episodic | paper · data | 47 conv. | data_public |
| Stateful SWE-bench | 10 | coding_cross_session, personalization | paper · code · data | 15 conv. 500 Q | official_script |
| TeleEgo | 10 | episodic, multimodal | paper · code · data | 5 conv. 3,291 Q | official_script |
| FlashAdventure | 09 | episodic, multimodal | paper · code | 34 Q | official_script |
| ZH-4O | 09 | conversational_longterm, personalization | paper · code | 28 conv. 1,068 Q | official_script |
| M3-Bench | 08 | episodic, multimodal | paper · code · data | 1,020 conv. 4,490 Q | official_script |
| MPR | 08 | personalization, parametric | paper · code · data | 12 conv. 108,000 Q | official_script |
| OdysseyBench | 08 | conversational_longterm | paper · code | 602 conv. 602 Q | official_script |
| StuLife | 08 | episodic, procedural_workflow | paper · code · data | 1,284 Q | official_script |
| LOCCO | 07 | conversational_longterm, parametric | paper · code | 100 conv. 2,981 Q | data_public |
| MemoryAgentBench | 07 | conversational_longterm, episodic, semantic_knowledge_update | paper · code · data | 2,071 Q | official_script |
| EvaLearn | 06 | procedural_workflow | paper · code | 182 conv. 648 Q | official_script |
| MemBench | 06 | conversational_longterm, personalization, semantic_knowledge_update | paper · code · data | 820 conv. 820 Q | data_public |
| OpenUnlearning | 06 | parametric, memory_security | paper · code · data | 400 Q | official_script |
| StoryBench | 06 | episodic | paper | 86 Q | paper_only |
| SWE-Bench-CL | 06 | coding_cross_session, procedural_workflow | paper · code · data | 8 conv. 273 Q | official_script |
| LifelongAgentBench | 05 | procedural_workflow | paper · code · data | 1,396 Q | official_script |
| MEMENTO | 05 | personalization, episodic, embodied | paper · code | 438 Q | official_script |
| MS-TOD | 05 | conversational_longterm | paper · code | 132 conv. 956 Q | official_script |
| UniEdit | 05 | parametric | paper · code · data | 311,142 Q | official_script |
| WikiDYK | 05 | parametric | paper · code · data | 77,180 Q | official_script |
| KUP (Knowledge Update Playground) | 04 | parametric | paper · code · data | 6,260 Q | official_script |
| PersonaMem | 04 | personalization, conversational_longterm, semantic_knowledge_update | paper · code · data | 60 conv. 2,727 Q | official_script |
| TiC-LM | 04 | parametric | paper · code | — | official_script |
| ImplexConv | 03 | conversational_longterm, personalization | paper · code · data | 2,500 conv. | data_public |
| LifeState-Bench | 03 | episodic, parametric, multi_agent_group, semantic_knowledge_update | paper | 311 Q | paper_only |
| WikiBigEdit | 03 | parametric | paper · code · data | 502,382 Q | official_script |
| MemoryCode | 02 | coding_cross_session, conversational_longterm, semantic_knowledge_update | paper · code · data | 360 conv. | official_script |
| Minerva | 02 | semantic_knowledge_update | paper · code | 1,110 Q | official_script |
| PersonaBench | 02 | personalization, conversational_longterm, semantic_knowledge_update | paper · code | 15 conv. 582 Q | official_script |
| PrefEval | 02 | personalization, conversational_longterm | paper · code · data | 3,000 Q | official_script |
| QAEdit | 02 | parametric | paper · code · data | 19,249 Q | official_script |
| REALTALK | 02 | conversational_longterm, episodic | paper · code | 10 conv. 728 Q | official_script |
| TReMu | 02 | conversational_longterm, episodic | paper | 600 Q | paper_only |
| Episodic Memory Benchmark | 01 | episodic, parametric | paper · code · data | 686 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| EvoWiki | 12 | semantic_knowledge_update, parametric | paper · code · data | 10,264 Q | paper_only |
| HalluEditBench | 10 | parametric | paper · code · data | 6,738 Q | official_script |
| LongMemEval | 10 | conversational_longterm, personalization, semantic_knowledge_update, episodic | paper · code · data | 500 conv. 500 Q | official_script |
| GoodAI LTM Benchmark | 09 | conversational_longterm, semantic_knowledge_update, episodic | paper · code | 1 conv. 33 Q | official_script |
| MADial-Bench | 09 | conversational_longterm, episodic | paper · code | 160 conv. | official_script |
| MemDaily | 09 | personalization | paper · code | 2,954 conv. 2,954 Q | official_script |
| CodeUpdateArena | 07 | parametric | paper · code · data | 670 Q | official_script |
| DialSim | 06 | conversational_longterm, multi_agent_group, episodic | paper · code · data | 3 conv. | official_script |
| StreamBench | 06 | procedural_workflow | paper · code · data | 9,702 Q | official_script |
| TeaFarm | 06 | conversational_longterm | paper · code · data | 100 conv. 200 Q | data_public |
| LoCoMo | 02 | conversational_longterm, episodic | paper · code | 10 conv. 1,986 Q | official_script |
| PerLTQA | 02 | personalization, episodic, conversational_longterm | paper · code | 3,409 conv. 8,593 Q | data_public |
| KnowEdit | 01 | parametric | paper · code · data | 5,653 Q | official_script |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| AToKe | 12 | parametric | paper · code | 8,820 Q | data_public |
| ConflictEdit / RoundEdit | 10 | parametric | paper · code | 2,500 Q | official_script |
| DMR (Deep Memory Retrieval) | 10 | conversational_longterm | paper · code · data | 500 conv. 500 Q | data_public |
| MMEdit | 10 | parametric, multimodal | paper · code · data | 3,093 Q | official_script |
| RippleEdits | 07 | parametric | paper · code | 4,000 Q | official_script |
| MemoryBank | 05 | conversational_longterm, episodic | paper · code | 15 conv. 194 Q | data_public |
| MQuAKE | 05 | parametric, semantic_knowledge_update | paper · code | 3,000 Q | official_script |
| MemoEval (SCM Self-Controlled Memory evaluation dataset) | 04 | conversational_longterm | paper · code | 18 conv. 105 Q | data_public |
| Name | Month | Types | Links | Scale | Repro level |
|---|---|---|---|---|---|
| Multi-Session Chat (MSC) | 07 | conversational_longterm, personalization | paper · code · data | 501 conv. | official_script |
| Level | Meaning |
|---|---|
unverified | No one has opened the paper and repository to check yet |
paper_only | Paper checked; no public data or code found |
data_public | Data is publicly available, but the official repository has no evaluation script |
official_script | The official repository has a script that runs the evaluation, and the data is available |
smoke_tested | We ran the official script on a small sample |
full_run | We ran the full evaluation |
benchmarks/<slug>.yaml following docs/fields.md, and cite a source in sources for every non-null field.python scripts/validate.py and python scripts/build_readme.py.benchmarks/, docs/, schema/ and the READMEs): CC BY 4.0scripts/, templates/, site/, pending/tools/): MITPapers, code and datasets linked from the entries are third-party works under their own licenses, which this repository's licenses do not cover. Check the original source before use.