--- title: "Entities · minds and possible minds, sorted by evidence · Simulation Hypothesis" url: https://simulation-hypothesis.com/entities description: "A sourced guide to minds: animals with measured neuron counts, collective and artificial systems, and the published tests for sentience and consciousness." license: https://creativecommons.org/licenses/by/4.0/ credit: "Ross Allan Dahl, Simulation Hypothesis (simulation-hypothesis.com)" --- # Entities · minds and possible minds, sorted by evidence If the world is a simulation, minds may be some of the hardest things in it to compute. This guide lists the known cases, from single cells to humans. It also covers the artificial systems built so far, and the minds that scientists and philosophers have proposed but nobody has observed. Every figure has a source and a confidence tag. measured is a direct count, estimate is worked out from other data, and not counted means nobody has published a count. Whether any of these systems has experiences is a separate question. The evidence for that is under [sentience criteria](https://simulation-hypothesis.com/entities/sentience) and [consciousness indicators](https://simulation-hypothesis.com/entities/indicators). ## Individual minds ## Animals: neurons in the cortex and the whole brain Cortex or pallium Whole brain All Mammals Birds Neurons in the cerebral cortex (mammals) or the pallium (birds), a widely used anatomical guide to cognitive capacity. Logarithmic scale; solid bars are counts, hatched bars estimates or disputed counts (†). Reptiles and invertebrates are not shown, as no comparable count is available. 1. Orca † 43.1 billion 2. Long-finned pilot whale † 37.2 billion 3. [Human](https://simulation-hypothesis.com/entities/human)▲1 16.3 billion 4. Harbour porpoise † 14.9 billion 5. [Bottlenose dolphin †](https://simulation-hypothesis.com/entities/dolphin) 12.7 billion 6. Cuvier’s beaked whale 9.1 billion 7. Gorilla 8.5 billion 8. Orangutan 8.5 billion 9. [Chimpanzee](https://simulation-hypothesis.com/entities/chimp) 7.4 billion 10. Common dolphin 6.7 billion 11. Harp seal 6.1 billion 12. [African elephant](https://simulation-hypothesis.com/entities/elephant)▼1 5.6 billion 13. Fin whale 5.2 billion 14. Blue whale 5.1 billion 15. Minke whale 3.2 billion Show all 88 10^6 10^7 10^8 10^9 10^10 10^11 neurons Show the numbers | # | Animal | Cortex or pallium | Whole brain | | -- | ------------------------- | ----------------- | ------------ | | 1 | Orca † | 43.1 billion | | | 2 | Long-finned pilot whale † | 37.2 billion | | | 3 | Human | 16.3 billion | 86.1 billion | | 4 | Harbour porpoise † | 14.9 billion | | | 5 | Bottlenose dolphin † | 12.7 billion | | | 6 | Cuvier’s beaked whale | 9.1 billion | | | 7 | Gorilla | 8.5 billion | | | 8 | Orangutan | 8.5 billion | | | 9 | Chimpanzee | 7.4 billion | | | 10 | Common dolphin | 6.7 billion | | | 11 | Harp seal | 6.1 billion | | | 12 | African elephant | 5.6 billion | 257 billion | | 13 | Fin whale | 5.2 billion | | | 14 | Blue whale | 5.1 billion | | | 15 | Minke whale | 3.2 billion | 57.4 billion | | 16 | Olive baboon | 2.88 billion | 10.9 billion | | 17 | Beluga | 2.7 billion | | | 18 | Pygmy sperm whale | 2.02 billion | | | 19 | Blue-and-yellow macaw | 1.92 billion | 3.14 billion | | 20 | Rhesus macaque | 1.71 billion | 6.39 billion | | 21 | Giraffe | 1.67 billion | 11.2 billion | | 22 | Squirrel monkey | 1.34 billion | 3.25 billion | | 23 | Eurasian eagle-owl | 1.33 billion | 1.9 billion | | 24 | Kea | 1.28 billion | 2.15 billion | | 25 | Snowy owl | 1.27 billion | 1.81 billion | | 26 | Common raven | 1.2 billion | 2.17 billion | | 27 | Tawny owl | 1.15 billion | 1.58 billion | | 28 | Capuchin monkey | 1.14 billion | 3.69 billion | | 29 | Sulphur-crested cockatoo | 1.14 billion | 2.12 billion | | 30 | African grey parrot | 850 million | 1.57 billion | | 31 | Rook | 820 million | 1.51 billion | | 32 | Greater kudu | 793 million | 6.09 billion | | 33 | Long-eared owl | 673 million | 991 million | | 34 | Goffin's cockatoo | 599 million | 1.16 billion | | 35 | Lion | 545 million | 4.67 billion | | 36 | White-tailed eagle | 543 million | 1.65 billion | | 37 | Eurasian jay | 529 million | 1.09 billion | | 38 | Domestic dog | 526 million | | | 39 | Striped hyena | 495 million | 3.88 billion | | 40 | Western jackdaw | 492 million | 968 million | | 41 | Common ostrich | 479 million | 1.63 billion | | 42 | Eurasian magpie | 443 million | 897 million | | 43 | Saker falcon | 443 million | 1.2 billion | | 44 | Emu | 439 million | 1.34 billion | | 45 | Raccoon | 438 million | 2.15 billion | | 46 | Barn owl | 437 million | 690 million | | 47 | Little owl | 401 million | 586 million | | 48 | Black vulture | 399 million | 1.01 billion | | 49 | Eurasian pygmy owl | 364 million | 533 million | | 50 | Common buzzard | 352 million | 1 billion | | 51 | Greater rhea | 330 million | 1.03 billion | | 52 | Mute swan | 324 million | 1 billion | | 53 | Domestic pig | 293 million | 2.22 billion | | 54 | Greylag goose | 259 million | 738 million | | 55 | Brown bear | 251 million | 9.59 billion | | 56 | Domestic cat | 250 million | 1.22 billion | | 57 | Common marmoset | 245 million | 638 million | | 58 | Common kestrel | 238 million | 625 million | | 59 | Eurasian teal | 167 million | 375 million | | 60 | Budgerigar | 149 million | 322 million | | 61 | Red-breasted goose | 149 million | 483 million | | 62 | Wood duck | 138 million | 306 million | | 63 | Pygmy falcon | 132 million | 350 million | | 64 | Marbled duck | 130 million | 387 million | | 65 | Indian peafowl | 130 million | 571 million | | 66 | Yellow-knobbed curassow | 125 million | 653 million | | 67 | Victoria crowned pigeon | 118 million | 579 million | | 68 | Mallard | 112 million | 367 million | | 69 | Wild turkey | 106 million | 493 million | | 70 | Eurasian sparrowhawk | 87.8 million | 402 million | | 71 | Reeves’s pheasant | 80.7 million | 387 million | | 72 | Rock pigeon | 72 million | 310 million | | 73 | European rabbit | 71.5 million | 513 million | | 74 | Red junglefowl | 60.7 million | 221 million | | 75 | Chilean tinamou | 59.1 million | 241 million | | 76 | Zebra finch | 55.2 million | 136 million | | 77 | Elegant crested tinamou | 51.4 million | 218 million | | 78 | Common wood pigeon | 51.3 million | 259 million | | 79 | Quebracho crested tinamou | 48.3 million | 236 million | | 80 | African collared dove | 40 million | 139 million | | 81 | Northern bobwhite | 39.1 million | 148 million | | 82 | Grey partridge | 36.9 million | 170 million | | 83 | Rat | 31 million | 200 million | | 84 | Common quail | 22.6 million | 118 million | | 85 | King quail | 20.5 million | 80.5 million | | 86 | Diamond dove | 18.2 million | 87.9 million | | 87 | Mouse | 13.7 million | 70.9 million | | 88 | Naked mole-rat | 5.31 million | 33.6 million | ## AI systems: parameters against synapses All parameters Used per token All Language Multimodal Vision Other (speech, biology, games, robotics…) Trainable parameters (connection weights), on a logarithmic scale: every notable model in Epoch AI’s database with a parameter count (retrieved 3 October 2026). Solid bars are counts Epoch rates confident; hatched bars are counts it rates likely or speculative. Bars marked \* are estimates for closed models whose developers do not publish their size, compiled 2 October 2026 from third-party analyses; the whisker shows the estimate’s range. Gemini 3 and later: no published estimate. Hollow rows are synapse counts: a parameter is closer to a synapse than to a neuron. 1. Human neocortex 1.6 × 10¹⁴ 2. GPT-5.5 (OpenAI, 2026) \* \~4.7 × 10¹² 3. Claude Fable 5 (Anthropic, 2026) \* \~3.5 × 10¹² 4. Grok 3 (xAI, 2025) \~3 × 10¹² 5. Grok 4 (xAI, 2025) \~3 × 10¹² 6. Gemini 2.5 Pro (Google, 2025) \* \~3 × 10¹² 7. Kimi K3 (Moonshot, 2026) \~2.8 × 10¹² 8. Qwen 3.8 Max (Alibaba, 2026) 2.4 × 10¹² 9. Qwen3.8-2.4T-A95B (Alibaba, 2026) 2.4 × 10¹² 10. Grok 4.7 (xAI, 2026) \* \~2.1 × 10¹² 11. Llama 4 Behemoth (Meta AI, 2025) \~2 × 10¹² 12. GPT-4 (OpenAI, 2023) \~1.8 × 10¹² 13. DeepSeek-V4-Pro (DeepSeek, 2026) \~1.6 × 10¹² 14. LongCat-2.0 (Meituan Inc, 2026) \~1.6 × 10¹² 15. Switch (Google, 2021) 1.6 × 10¹² 16. GLaM (Google, 2021) 1.2 × 10¹² 17. PanGu-Σ (Huawei Noah's Ark Lab, 2023) 1.1 × 10¹² 18. Kimi K2.5 (Moonshot, 2026) \~1 × 10¹² 19. Composer 2 (Cursor, 2026) 1 × 10¹² 20. Kimi K2.6 (Moonshot, 2026) 1 × 10¹² 21. Composer 2.5 (Cursor, 2026) \~1 × 10¹² 22. MiMo-V2-Pro (Xiaomi Corp, 2026) 1 × 10¹² 23. MiMo-V2.5-Pro (Xiaomi Corp, 2026) \~1 × 10¹² 24. M6-T (Alibaba, 2021) \~1 × 10¹² 25. Kimi K2 (Moonshot, 2025) 1 × 10¹² 26. Qwen3-Max (Alibaba, 2025) \~1 × 10¹² 27. Ling-1T (Ant Group, 2025) 1 × 10¹² 28. Kimi K2 Thinking (Moonshot, 2025) \~1 × 10¹² 29. Inkling (Thinking Machines, 2026) \~9.8 × 10¹¹ 30. Tencent Hy4 preview (Tencent, 2026) \~7.7 × 10¹¹ 31. GLM-5.1 (Z.ai (Zhipu AI), 2026) 7.5 × 10¹¹ 32. K-EXAONE 2.0 (LG AI Research, 2026) 7.5 × 10¹¹ 33. DeepSeek V4.1 Flash (DeepSeek, 2026) \~7.5 × 10¹¹ 34. GLM-5 (Z.ai (Zhipu AI), 2026) \~7.4 × 10¹¹ 35. GLM-5.3 (Z.ai (Zhipu AI), 2026) 7.4 × 10¹¹ 36. A.X K2 (SK Telecom, 2026) 6.9 × 10¹¹ 37. DeepSeekMath-V2 (DeepSeek, 2025) 6.8 × 10¹¹ 38. Mistral Large 3 (Mistral AI, 2025) 6.8 × 10¹¹ 39. DeepSeek-V3 (DeepSeek, 2024) 6.7 × 10¹¹ 40. DeepSeek-R1 (DeepSeek, 2025) 6.7 × 10¹¹ Show all 725 10^4 10^6 10^8 10^10 10^12 10^14 parameters or synapses Show the numbers | # | System | All parameters | Used per token | Year | Kind | | --- | ------------------------------------------------------------------------------------------- | -------------- | -------------- | ---- | --------------------------------------------- | | 1 | Human neocortex | 1.6 × 10¹⁴ | 1.6 × 10¹⁴ | | synapses | | 2 | GPT-5.5 (OpenAI, 2026) \* | \~4.7 × 10¹² | | 2026 | estimate | | 3 | Claude Fable 5 (Anthropic, 2026) \* | \~3.5 × 10¹² | | 2026 | estimate | | 4 | Grok 3 (xAI, 2025) | \~3 × 10¹² | | 2025 | estimate | | 5 | Grok 4 (xAI, 2025) | \~3 × 10¹² | | 2025 | estimate | | 6 | Gemini 2.5 Pro (Google, 2025) \* | \~3 × 10¹² | | 2025 | estimate | | 7 | Kimi K3 (Moonshot, 2026) | \~2.8 × 10¹² | | 2026 | estimate | | 8 | Qwen 3.8 Max (Alibaba, 2026) | 2.4 × 10¹² | 2.4 × 10¹² | 2026 | architecture not stated | | 9 | Qwen3.8-2.4T-A95B (Alibaba, 2026) | 2.4 × 10¹² | 9.5 × 10¹⁰ | 2026 | mixture of experts | | 10 | Grok 4.7 (xAI, 2026) \* | \~2.1 × 10¹² | | 2026 | estimate | | 11 | Llama 4 Behemoth (Meta AI, 2025) | \~2 × 10¹² | | 2025 | estimate | | 12 | GPT-4 (OpenAI, 2023) | \~1.8 × 10¹² | | 2023 | estimate | | 13 | DeepSeek-V4-Pro (DeepSeek, 2026) | \~1.6 × 10¹² | | 2026 | estimate | | 14 | LongCat-2.0 (Meituan Inc, 2026) | \~1.6 × 10¹² | | 2026 | estimate | | 15 | Switch (Google, 2021) | 1.6 × 10¹² | 1.6 × 10¹² | 2021 | mixture of experts (active count unpublished) | | 16 | GLaM (Google, 2021) | 1.2 × 10¹² | 1.2 × 10¹² | 2021 | mixture of experts (active count unpublished) | | 17 | PanGu-Σ (Huawei Noah's Ark Lab, 2023) | 1.1 × 10¹² | 1.1 × 10¹² | 2023 | mixture of experts (active count unpublished) | | 18 | Kimi K2.5 (Moonshot, 2026) | \~1 × 10¹² | | 2026 | estimate | | 19 | Composer 2 (Cursor, 2026) | 1 × 10¹² | 1 × 10¹² | 2026 | mixture of experts (active count unpublished) | | 20 | Kimi K2.6 (Moonshot, 2026) | 1 × 10¹² | 1 × 10¹² | 2026 | mixture of experts (active count unpublished) | | 21 | Composer 2.5 (Cursor, 2026) | \~1 × 10¹² | | 2026 | estimate | | 22 | MiMo-V2-Pro (Xiaomi Corp, 2026) | 1 × 10¹² | 1 × 10¹² | 2026 | architecture not stated | | 23 | MiMo-V2.5-Pro (Xiaomi Corp, 2026) | \~1 × 10¹² | | 2026 | estimate | | 24 | M6-T (Alibaba, 2021) | \~1 × 10¹² | | 2021 | estimate | | 25 | Kimi K2 (Moonshot, 2025) | 1 × 10¹² | 1 × 10¹² | 2025 | mixture of experts (active count unpublished) | | 26 | Qwen3-Max (Alibaba, 2025) | \~1 × 10¹² | | 2025 | estimate | | 27 | Ling-1T (Ant Group, 2025) | 1 × 10¹² | 5 × 10¹⁰ | 2025 | mixture of experts | | 28 | Kimi K2 Thinking (Moonshot, 2025) | \~1 × 10¹² | | 2025 | estimate | | 29 | Inkling (Thinking Machines, 2026) | \~9.8 × 10¹¹ | | 2026 | estimate | | 30 | Tencent Hy4 preview (Tencent, 2026) | \~7.7 × 10¹¹ | | 2026 | estimate | | 31 | GLM-5.1 (Z.ai (Zhipu AI), 2026) | 7.5 × 10¹¹ | 7.5 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 32 | K-EXAONE 2.0 (LG AI Research, 2026) | 7.5 × 10¹¹ | 7.5 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 33 | DeepSeek V4.1 Flash (DeepSeek, 2026) | \~7.5 × 10¹¹ | | 2026 | estimate | | 34 | GLM-5 (Z.ai (Zhipu AI), 2026) | \~7.4 × 10¹¹ | | 2026 | estimate | | 35 | GLM-5.3 (Z.ai (Zhipu AI), 2026) | 7.4 × 10¹¹ | 4 × 10¹⁰ | 2026 | mixture of experts | | 36 | A.X K2 (SK Telecom, 2026) | 6.9 × 10¹¹ | 6.9 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 37 | DeepSeekMath-V2 (DeepSeek, 2025) | 6.8 × 10¹¹ | 6.8 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 38 | Mistral Large 3 (Mistral AI, 2025) | 6.8 × 10¹¹ | 4.1 × 10¹⁰ | 2025 | mixture of experts | | 39 | DeepSeek-V3 (DeepSeek, 2024) | 6.7 × 10¹¹ | 3.7 × 10¹⁰ | 2024 | mixture of experts | | 40 | DeepSeek-R1 (DeepSeek, 2025) | 6.7 × 10¹¹ | 3.7 × 10¹⁰ | 2025 | mixture of experts | | 41 | SenseChat 5.5 (SenseTime, 2024) | 6 × 10¹¹ | 6 × 10¹¹ | 2024 | architecture not stated | | 42 | Step 5 Preview (StepFun, 2026) | \~6 × 10¹¹ | | 2026 | estimate | | 43 | PaLM-E (Google, 2023) | \~5.6 × 10¹¹ | | 2023 | estimate | | 44 | Hunyuan-TurboS (Tencent, 2025) | 5.6 × 10¹¹ | 5.6 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 45 | LongCat-Flash (Meituan Inc, 2025) | \~5.6 × 10¹¹ | | 2025 | estimate | | 46 | Nemotron 3 Ultra (Nvidia, 2026) | \~5.5 × 10¹¹ | | 2026 | estimate | | 47 | PaLM (Google Research, 2022) | 5.4 × 10¹¹ | 5.4 × 10¹¹ | 2022 | dense | | 48 | Minerva (Google, 2022) | 5.4 × 10¹¹ | 5.4 × 10¹¹ | 2022 | dense | | 49 | U-PaLM (Google, 2022) | 5.4 × 10¹¹ | 5.4 × 10¹¹ | 2022 | dense | | 50 | LMSI-Palm (Google, 2022) | 5.4 × 10¹¹ | 5.4 × 10¹¹ | 2022 | dense | | 51 | Flan-PaLM 540B (Google, 2022) | 5.4 × 10¹¹ | 5.4 × 10¹¹ | 2022 | dense | | 52 | Megatron-Turing NLG 530B (Microsoft, 2021) | 5.3 × 10¹¹ | 5.3 × 10¹¹ | 2021 | dense | | 53 | MegaScale (ByteDance, 2024) | \~5.3 × 10¹¹ | | 2024 | estimate | | 54 | A.X K1 (SK Telecom, 2025) | 5.2 × 10¹¹ | 5.2 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 55 | Doubao-pro (ByteDance, 2024) | \~5 × 10¹¹ | | 2024 | estimate | | 56 | Grok 4.20 (xAI, 2026) | \~5 × 10¹¹ | | 2026 | estimate | | 57 | Qwen3-Coder-480B-A35B (Alibaba, 2025) | 4.8 × 10¹¹ | 3.5 × 10¹⁰ | 2025 | mixture of experts | | 58 | MiniMax-M3 (MiniMax, 2026) | 4.3 × 10¹¹ | 2.2 × 10¹⁰ | 2026 | mixture of experts | | 59 | ERNIE-4.5-VL-424B-A47B (Baidu, 2025) | 4.2 × 10¹¹ | 4.7 × 10¹⁰ | 2025 | mixture of experts | | 60 | Llama 3.1-405B (Meta AI, 2024) | 4 × 10¹¹ | 4 × 10¹¹ | 2024 | dense | | 61 | Llama 4 Maverick (Meta AI, 2025) | \~4 × 10¹¹ | | 2025 | estimate | | 62 | Jamba 1.5-Large (AI21 Labs, 2024) | 4 × 10¹¹ | 4 × 10¹¹ | 2024 | mixture of experts (active count unpublished) | | 63 | Qwen3.5 397B-A17B (Alibaba, 2026) | \~4 × 10¹¹ | | 2026 | estimate | | 64 | Hunyuan-Large (Tencent, 2024) | 3.9 × 10¹¹ | 3.9 × 10¹¹ | 2024 | mixture of experts (active count unpublished) | | 65 | GLM-4.7 (Z.ai (Zhipu AI), 2025) | \~3.6 × 10¹¹ | | 2025 | estimate | | 66 | GLM-4.6 (Z.ai (Zhipu AI), 2025) | \~3.6 × 10¹¹ | | 2025 | estimate | | 67 | GLM-4.5 (Z.ai (Zhipu AI), 2025) | 3.5 × 10¹¹ | 3.2 × 10¹⁰ | 2025 | mixture of experts | | 68 | PaLM 2 (Google, 2023) | \~3.4 × 10¹¹ | | 2023 | estimate | | 69 | Med-PaLM 2 (Google Research, 2023) | \~3.4 × 10¹¹ | | 2023 | estimate | | 70 | Nemotron-4 340B (Nvidia, 2024) | 3.4 × 10¹¹ | 3.4 × 10¹¹ | 2024 | dense | | 71 | GLM-5.3-Flash (Z.ai (Zhipu AI), 2026) | \~3.2 × 10¹¹ | | 2026 | estimate | | 72 | Grok-1 (xAI, 2023) | \~3.1 × 10¹¹ | | 2023 | estimate | | 73 | Motif-3 (Motif Technologies, 2026) | 3.1 × 10¹¹ | 3.1 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 74 | SB-LM (Google, 2007) | \~3 × 10¹¹ | | 2007 | estimate | | 75 | Tencent Hy3 preview (Tencent, 2026) | 3 × 10¹¹ | 3 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 76 | Tencent Hy3 (Tencent, 2026) | 3 × 10¹¹ | 3 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 77 | DeepSeek-V4-Flash (DeepSeek, 2026) | \~2.8 × 10¹¹ | | 2026 | estimate | | 78 | Gopher (DeepMind, 2021) | 2.8 × 10¹¹ | 2.8 × 10¹¹ | 2021 | dense | | 79 | TML-Interaction-Small (Thinking Machines, 2026) | 2.8 × 10¹¹ | 2.8 × 10¹¹ | 2026 | mixture of experts (active count unpublished) | | 80 | ST-MoE (Google, 2022) | \~2.7 × 10¹¹ | | 2022 | estimate | | 81 | ERNIE 3.0 Titan (Baidu, 2021) | 2.6 × 10¹¹ | 2.6 × 10¹¹ | 2021 | dense | | 82 | Solar Open2 250B (Upstage, 2026) | \~2.5 × 10¹¹ | | 2026 | estimate | | 83 | Aramco Metabrain AI (Saudi Aramco, 2024) | \~2.5 × 10¹¹ | | 2024 | estimate | | 84 | Yuan 1.0 (Inspur, 2021) | 2.5 × 10¹¹ | 2.5 × 10¹¹ | 2021 | architecture not stated | | 85 | DeepSeek-Coder-V2 236B (DeepSeek, 2024) | 2.4 × 10¹¹ | 2.4 × 10¹¹ | 2024 | mixture of experts (active count unpublished) | | 86 | DeepSeek-V2.5 (DeepSeek, 2024) | 2.4 × 10¹¹ | 2.4 × 10¹¹ | 2024 | mixture of experts (active count unpublished) | | 87 | K-EXAONE (LG AI Research, 2025) | 2.4 × 10¹¹ | 2.4 × 10¹¹ | 2025 | architecture not stated | | 88 | Qwen3-235B-A22B (Alibaba, 2025) | \~2.4 × 10¹¹ | | 2025 | estimate | | 89 | Qwen3-235B-A22B-Thinking (Alibaba, 2025) | \~2.4 × 10¹¹ | | 2025 | estimate | | 90 | P1-235B-A22B (Shanghai AI Lab, 2025) | \~2.4 × 10¹¹ | | 2025 | estimate | | 91 | Seed-1.6-Thinking (ByteDance, 2025) | 2.3 × 10¹¹ | 2.3 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 92 | MiniMax-M2 (MiniMax, 2025) | 2.3 × 10¹¹ | 1 × 10¹⁰ | 2025 | mixture of experts | | 93 | MiniMax-M2.1 (MiniMax, 2025) | 2.3 × 10¹¹ | 2.3 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 94 | HyperCLOVA 204B (NAVER, 2021) | \~2 × 10¹¹ | | 2021 | estimate | | 95 | Amazon Titan (Amazon, 2023) | \~2 × 10¹¹ | | 2023 | estimate | | 96 | Falcon-180B (Technology Innovation Institute, 2023) | 1.8 × 10¹¹ | 1.8 × 10¹¹ | 2023 | dense | | 97 | Jurassic-1-Jumbo (AI21 Labs, 2021) | 1.8 × 10¹¹ | 1.8 × 10¹¹ | 2021 | architecture not stated | | 98 | BLOOM-176B (Hugging Face, 2022) | 1.8 × 10¹¹ | 1.8 × 10¹¹ | 2022 | dense | | 99 | BLUUMI (University of Turku, 2023) | \~1.8 × 10¹¹ | | 2023 | estimate | | 100 | InstructGPT 175B (OpenAI, 2022) | 1.8 × 10¹¹ | 1.8 × 10¹¹ | 2022 | dense | | 101 | OPT-175B (Meta AI, 2022) | 1.8 × 10¹¹ | 1.8 × 10¹¹ | 2022 | dense | | 102 | BlenderBot 3 (McGill University, 2022) | \~1.8 × 10¹¹ | | 2022 | estimate | | 103 | GPT-3 175B (OpenAI, 2020) | 1.7 × 10¹¹ | 1.7 × 10¹¹ | 2020 | dense | | 104 | Palmyra X 004 (Writer, 2024) | \~1.5 × 10¹¹ | | 2024 | estimate | | 105 | FLAN 137B (Google Research, 2021) | 1.4 × 10¹¹ | 1.4 × 10¹¹ | 2021 | architecture not stated | | 106 | LaMDA (Google, 2022) | 1.4 × 10¹¹ | 1.4 × 10¹¹ | 2022 | architecture not stated | | 107 | MEB (Microsoft, 2021) | 1.4 × 10¹¹ | 1.4 × 10¹¹ | 2021 | architecture not stated | | 108 | Pangu Ultra (Huawei, 2025) | 1.4 × 10¹¹ | 1.4 × 10¹¹ | 2025 | architecture not stated | | 109 | DBRX (Databricks, 2024) | 1.3 × 10¹¹ | 1.3 × 10¹¹ | 2024 | mixture of experts (active count unpublished) | | 110 | GLM-130B (Tsinghua University, 2022) | 1.3 × 10¹¹ | 1.3 × 10¹¹ | 2022 | dense | | 111 | Pixtral Large (Mistral AI, 2024) | 1.2 × 10¹¹ | 1.2 × 10¹¹ | 2024 | architecture not stated | | 112 | Mistral Large 2 (Mistral AI, 2024) | \~1.2 × 10¹¹ | | 2024 | estimate | | 113 | Qwen3.5-122B-A10B (Alibaba, 2026) | \~1.2 × 10¹¹ | | 2026 | estimate | | 114 | Galactica (Meta AI, 2022) | \~1.2 × 10¹¹ | | 2022 | estimate | | 115 | Nemotron 3 Super (Nvidia, 2026) | \~1.2 × 10¹¹ | | 2026 | estimate | | 116 | gpt-oss-120b (OpenAI, 2025) | 1.2 × 10¹¹ | 5.1 × 10⁹ | 2025 | mixture of experts | | 117 | Telechat2-115B (China Telecom, 2024) | 1.1 × 10¹¹ | 1.1 × 10¹¹ | 2024 | architecture not stated | | 118 | Kaiju Large (Character.ai, 2025) | \~1.1 × 10¹¹ | | 2025 | estimate | | 119 | Llama 4 Scout (Meta AI, 2025) | \~1.1 × 10¹¹ | | 2025 | estimate | | 120 | InternLM (Shanghai AI Lab, 2023) | 1 × 10¹¹ | 1 × 10¹¹ | 2023 | dense | | 121 | Solar Open 100B (Upstage, 2025) | 1 × 10¹¹ | 1 × 10¹¹ | 2025 | architecture not stated | | 122 | DLRM-2020 (Facebook AI, 2019) | 1 × 10¹¹ | 1 × 10¹¹ | 2019 | dense | | 123 | xTrimoPGLM -100B (Tsinghua University, 2023) | 1 × 10¹¹ | 1 × 10¹¹ | 2023 | architecture not stated | | 124 | Yi-Large (01.AI, 2024) | \~1 × 10¹¹ | | 2024 | estimate | | 125 | VAETKI (NC AI, 2025) | 1 × 10¹¹ | 1 × 10¹¹ | 2025 | mixture of experts (active count unpublished) | | 126 | ESM3 (EvolutionaryScale, 2024) | 9.8 × 10¹⁰ | 9.8 × 10¹⁰ | 2024 | architecture not stated | | 127 | Mouse neocortex | 8.1 × 10¹⁰ | 8.1 × 10¹⁰ | | synapses | | 128 | Flamingo (DeepMind, 2022) | 8 × 10¹⁰ | 8 × 10¹⁰ | 2022 | dense | | 129 | Qwen3-Coder-Next (Alibaba, 2026) | \~8 × 10¹⁰ | | 2026 | estimate | | 130 | Qwen2-72B (Alibaba, 2024) | 7.3 × 10¹⁰ | 7.3 × 10¹⁰ | 2024 | dense | | 131 | Qwen2.5-72B (Alibaba, 2024) | 7.3 × 10¹⁰ | 7.3 × 10¹⁰ | 2024 | dense | | 132 | Qwen2.5 Instruct (Alibaba, 2024) | 7.3 × 10¹⁰ | 7.3 × 10¹⁰ | 2024 | dense | | 133 | Qwen-72B (Alibaba, 2023) | 7.2 × 10¹⁰ | 7.2 × 10¹⁰ | 2023 | dense | | 134 | Palmyra X 003 (Writer, 2024) | \~7.2 × 10¹⁰ | | 2024 | estimate | | 135 | Qwen1.5-72B (Alibaba, 2024) | 7.2 × 10¹⁰ | 7.2 × 10¹⁰ | 2024 | dense | | 136 | LLaVA-OV-72B (ByteDance, 2024) | 7.2 × 10¹⁰ | 7.2 × 10¹⁰ | 2024 | dense | | 137 | NVLM-X 72B (Nvidia, 2024) | \~7.2 × 10¹⁰ | | 2024 | estimate | | 138 | NVLM-H 72B (Nvidia, 2024) | \~7.2 × 10¹⁰ | | 2024 | estimate | | 139 | NVLM-D 72B (Nvidia, 2024) | 7.2 × 10¹⁰ | 7.2 × 10¹⁰ | 2024 | architecture not stated | | 140 | MindLink-72B (Kunlun Inc., 2025) | 7.2 × 10¹⁰ | 7.2 × 10¹⁰ | 2025 | architecture not stated | | 141 | Chinchilla (DeepMind, 2022) | 7 × 10¹⁰ | 7 × 10¹⁰ | 2022 | dense | | 142 | Llama 2-70B (Meta AI, 2023) | 7 × 10¹⁰ | 7 × 10¹⁰ | 2023 | dense | | 143 | PPLX-70B-Online (Perplexity, 2023) | \~7 × 10¹⁰ | | 2023 | estimate | | 144 | Llama 3-70B (Meta AI, 2024) | 7 × 10¹⁰ | 7 × 10¹⁰ | 2024 | dense | | 145 | ALLaM adapted 70B (Saudi Data and Artificial Intelligence Authority, 2024) | 7 × 10¹⁰ | 7 × 10¹⁰ | 2024 | architecture not stated | | 146 | Llama 3.3 70B (Meta AI, 2024) | 7 × 10¹⁰ | 7 × 10¹⁰ | 2024 | dense | | 147 | Reka Core (Reka AI, 2024) | \~6.7 × 10¹⁰ | | 2024 | estimate | | 148 | LLaMA-65B (Meta AI, 2023) | 6.5 × 10¹⁰ | 6.5 × 10¹⁰ | 2023 | dense | | 149 | SNM-skip (Google, 2014) | \~6.2 × 10¹⁰ | | 2014 | estimate | | 150 | PaLI-X (Google Research, 2023) | \~5.5 × 10¹⁰ | | 2023 | estimate | | 151 | RT-2 (Google DeepMind, 2023) | 5.5 × 10¹⁰ | 5.5 × 10¹⁰ | 2023 | architecture not stated | | 152 | RT-2-X (Google DeepMind, 2023) | 5.5 × 10¹⁰ | 5.5 × 10¹⁰ | 2023 | architecture not stated | | 153 | NLLB (Meta AI, 2022) | 5.5 × 10¹⁰ | 5.5 × 10¹⁰ | 2022 | dense | | 154 | BloombergGPT (Bloomberg, 2023) | 5.1 × 10¹⁰ | 5.1 × 10¹⁰ | 2023 | dense | | 155 | M4-50B (Google, 2019) | 5 × 10¹⁰ | 5 × 10¹⁰ | 2019 | mixture of experts (active count unpublished) | | 156 | Mixtral 8x7B (Mistral AI, 2023) | \~4.7 × 10¹⁰ | | 2023 | estimate | | 157 | AlphaCode (DeepMind, 2022) | 4.1 × 10¹⁰ | 4.1 × 10¹⁰ | 2022 | architecture not stated | | 158 | Falcon-40B (Technology Innovation Institute, 2023) | 4 × 10¹⁰ | 4 × 10¹⁰ | 2023 | dense | | 159 | Qwen3-Omni-30B-A3B (Alibaba, 2025) | 3.5 × 10¹⁰ | 3 × 10⁹ | 2025 | mixture of experts | | 160 | Yi-34B (01.AI, 2023) | 3.4 × 10¹⁰ | 3.4 × 10¹⁰ | 2023 | dense | | 161 | Cambrian-1-34B (New York University (NYU), 2024) | 3.4 × 10¹⁰ | 3.4 × 10¹⁰ | 2024 | architecture not stated | | 162 | Oryx 34B (Tsinghua University, 2024) | 3.4 × 10¹⁰ | 3.4 × 10¹⁰ | 2024 | architecture not stated | | 163 | Kaiju Medium (Character.ai, 2025) | \~3.4 × 10¹⁰ | | 2025 | estimate | | 164 | EXAONE 4.5 (LG AI Research, 2026) | 3.3 × 10¹⁰ | 3.3 × 10¹⁰ | 2026 | architecture not stated | | 165 | Qwen2.5-32B (Alibaba, 2024) | 3.3 × 10¹⁰ | 3.3 × 10¹⁰ | 2024 | dense | | 166 | QwQ-32B (Alibaba, 2025) | \~3.3 × 10¹⁰ | | 2025 | estimate | | 167 | EXAONE 3.5 32B (LG AI Research, 2024) | 3.2 × 10¹⁰ | 3.2 × 10¹⁰ | 2024 | architecture not stated | | 168 | EXAONE Deep 32B (LG AI Research, 2025) | 3.2 × 10¹⁰ | 3.2 × 10¹⁰ | 2025 | architecture not stated | | 169 | EXAONE 4.0 (LG AI Research, 2025) | 3.2 × 10¹⁰ | 3.2 × 10¹⁰ | 2025 | architecture not stated | | 170 | Olmo 3 (Allen Institute for AI, 2025) | 3.2 × 10¹⁰ | 3.2 × 10¹⁰ | 2025 | architecture not stated | | 171 | HyperCLOVA X SEED 32B Think (NAVER, 2025) | 3.2 × 10¹⁰ | 3.2 × 10¹⁰ | 2025 | architecture not stated | | 172 | Nemotron 3-Nano-30B-A3B (Nvidia, 2025) | \~3.2 × 10¹⁰ | | 2025 | estimate | | 173 | Tongyi DeepResearch (Alibaba, 2025) | 3 × 10¹⁰ | 3 × 10¹⁰ | 2025 | architecture not stated | | 174 | MM1-30B (Apple, 2024) | \~3 × 10¹⁰ | | 2024 | estimate | | 175 | Movie Gen Video (Meta AI, 2024) | 3 × 10¹⁰ | 3 × 10¹⁰ | 2024 | architecture not stated | | 176 | AgentFounder-30B (Alibaba, 2025) | 3 × 10¹⁰ | 3 × 10¹⁰ | 2025 | architecture not stated | | 177 | PLUG (Alibaba, 2021) | 2.7 × 10¹⁰ | 2.7 × 10¹⁰ | 2021 | architecture not stated | | 178 | Qwen Image (Alibaba, 2025) | 2.7 × 10¹⁰ | 2.7 × 10¹⁰ | 2025 | architecture not stated | | 179 | GenSLM (University of Chicago, 2022) | 2.5 × 10¹⁰ | 2.5 × 10¹⁰ | 2022 | architecture not stated | | 180 | ViT-22B (Google, 2023) | 2.2 × 10¹⁰ | 2.2 × 10¹⁰ | 2023 | architecture not stated | | 181 | KN-LM (Google, 2007) | \~2.1 × 10¹⁰ | | 2007 | estimate | | 182 | gpt-oss-20b (OpenAI, 2025) | 2.1 × 10¹⁰ | 2.1 × 10¹⁰ | 2025 | mixture of experts (active count unpublished) | | 183 | RNN for 1B words (Google, 2013) | \~2 × 10¹⁰ | | 2013 | estimate | | 184 | GPT-NeoX-20B (EleutherAI, 2022) | 2 × 10¹⁰ | 2 × 10¹⁰ | 2022 | dense | | 185 | UL2 (Google Research, 2022) | 2 × 10¹⁰ | 2 × 10¹⁰ | 2022 | architecture not stated | | 186 | Parti (Google Research, 2022) | 2 × 10¹⁰ | 2 × 10¹⁰ | 2022 | architecture not stated | | 187 | GPT-3.5 Turbo (OpenAI, 2023) | \~2 × 10¹⁰ | | 2023 | estimate | | 188 | GPT-3.5 Turbo Instruct (OpenAI, 2023) | \~2 × 10¹⁰ | | 2023 | estimate | | 189 | SPHINX (Shanghai AI Lab, 2023) | \~2 × 10¹⁰ | | 2023 | estimate | | 190 | AlexaTM 20B (Amazon, 2022) | 2 × 10¹⁰ | 2 × 10¹⁰ | 2022 | architecture not stated | | 191 | CogAgent (Tsinghua University, 2023) | \~1.8 × 10¹⁰ | | 2023 | estimate | | 192 | Turing-NLG (Microsoft, 2020) | \~1.7 × 10¹⁰ | | 2020 | estimate | | 193 | CogVLM-17B (Tsinghua University, 2023) | 1.7 × 10¹⁰ | 1.7 × 10¹⁰ | 2023 | architecture not stated | | 194 | PaLI (Google, 2022) | \~1.7 × 10¹⁰ | | 2022 | estimate | | 195 | CodeT5+ (Salesforce, 2023) | 1.6 × 10¹⁰ | 1.6 × 10¹⁰ | 2023 | architecture not stated | | 196 | StarCoder (Hugging Face, 2023) | 1.6 × 10¹⁰ | 1.6 × 10¹⁰ | 2023 | dense | | 197 | ESM2-15B (Meta AI, 2022) | 1.5 × 10¹⁰ | 1.5 × 10¹⁰ | 2022 | architecture not stated | | 198 | FunSearch (Google DeepMind, 2023) | \~1.5 × 10¹⁰ | | 2023 | estimate | | 199 | NVILA 15B (Nvidia, 2024) | 1.5 × 10¹⁰ | 1.5 × 10¹⁰ | 2024 | architecture not stated | | 200 | nekomata-14b (rinna, 2023) | 1.4 × 10¹⁰ | 1.4 × 10¹⁰ | 2023 | architecture not stated | | 201 | VILA1.5-13B (Nvidia, 2024) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2024 | architecture not stated | | 202 | VILA-13B (Nvidia, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | architecture not stated | | 203 | mT5-XXL (Google, 2020) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2020 | dense | | 204 | mT0-13B (Hugging Face, 2022) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2022 | architecture not stated | | 205 | LLaVA (University of Wisconsin Madison, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | dense | | 206 | InstructBLIP (Salesforce Research, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | architecture not stated | | 207 | Jais (Cerebras Systems, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | architecture not stated | | 208 | FinGPT-13B (University of California Los Angeles (UCLA), 2023) | \~1.3 × 10¹⁰ | | 2023 | estimate | | 209 | Ferret (Columbia University, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | architecture not stated | | 210 | Skywork-13B (Kunlun Inc., 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | architecture not stated | | 211 | LLaVA 1.5 (University of Wisconsin Madison, 2023) | 1.3 × 10¹⁰ | 1.3 × 10¹⁰ | 2023 | dense | | 212 | Volcano 13B (Korea University, 2023) | \~1.3 × 10¹⁰ | | 2023 | estimate | | 213 | Aya (Cohere Labs (formerly Cohere for AI), 2024) | \~1.3 × 10¹⁰ | | 2024 | estimate | | 214 | Kaiju Small (Character.ai, 2025) | \~1.3 × 10¹⁰ | | 2025 | estimate | | 215 | ByT5-XXL (Google, 2021) | \~1.3 × 10¹⁰ | | 2021 | estimate | | 216 | CoRe (Tsinghua University, 2023) | \~1.2 × 10¹⁰ | | 2023 | estimate | | 217 | DALL-E (OpenAI, 2021) | \~1.2 × 10¹⁰ | | 2021 | estimate | | 218 | Codex (OpenAI, 2021) | \~1.2 × 10¹⁰ | | 2021 | estimate | | 219 | T5-11B (Google, 2019) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2019 | dense | | 220 | UnifiedQA (Allen Institute for AI, 2020) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2020 | dense | | 221 | Unicorn (Allen Institute for AI, 2021) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2021 | architecture not stated | | 222 | PLATO-XL (Baidu, 2021) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2021 | architecture not stated | | 223 | CoEdiT-xxl (University of Minnesota, 2023) | \~1.1 × 10¹⁰ | | 2023 | estimate | | 224 | GQA-8-XXL (Google Research, 2023) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2023 | architecture not stated | | 225 | XLMR-XXL (Facebook AI Research, 2021) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2021 | architecture not stated | | 226 | Llama 3.2 11B (Meta AI, 2024) | 1.1 × 10¹⁰ | 1.1 × 10¹⁰ | 2024 | dense | | 227 | ERNIE 3.0 (Baidu, 2021) | 1 × 10¹⁰ | 1 × 10¹⁰ | 2021 | dense | | 228 | ERNIE-ViLG (Baidu, 2021) | \~1 × 10¹⁰ | | 2021 | estimate | | 229 | Qwen-VL (Alibaba, 2023) | \~9.6 × 10⁹ | | 2023 | estimate | | 230 | Generative BST (Facebook AI Research, 2021) | 9.4 × 10⁹ | 9.4 × 10⁹ | 2021 | architecture not stated | | 231 | Sparse all-MLP (Meta AI, 2022) | 9.4 × 10⁹ | 9.4 × 10⁹ | 2022 | mixture of experts (active count unpublished) | | 232 | CogVideo (Tsinghua University, 2022) | \~9.4 × 10⁹ | | 2022 | estimate | | 233 | eDiff-I (Nvidia, 2022) | \~9.1 × 10⁹ | | 2022 | estimate | | 234 | Eagle 2 (Nvidia, 2025) | 8.9 × 10⁹ | 8.9 × 10⁹ | 2025 | architecture not stated | | 235 | MoE-Multi (Jagiellonian University, 2017) | 8.7 × 10⁹ | 8.7 × 10⁹ | 2017 | mixture of experts (active count unpublished) | | 236 | Qwen-Audio-Chat (Alibaba, 2023) | \~8.5 × 10⁹ | | 2023 | estimate | | 237 | Megatron-LM (Nvidia, 2019) | \~8.3 × 10⁹ | | 2019 | estimate | | 238 | BigSSL (Google, 2021) | \~8 × 10⁹ | | 2021 | estimate | | 239 | Nemotron-3-8B (Nvidia, 2023) | 8 × 10⁹ | 8 × 10⁹ | 2023 | architecture not stated | | 240 | Stable Diffusion 3 (Stability AI, 2024) | 8 × 10⁹ | 8 × 10⁹ | 2024 | dense | | 241 | Qwen3 Embedding (Alibaba, 2025) | 8 × 10⁹ | 8 × 10⁹ | 2025 | mixture of experts (active count unpublished) | | 242 | Imagen (Google Brain, 2022) | \~7.8 × 10⁹ | | 2022 | estimate | | 243 | XGLM-7.5B (Meta AI, 2021) | 7.5 × 10⁹ | 7.5 × 10⁹ | 2021 | architecture not stated | | 244 | RETRO-7B (DeepMind, 2022) | 7.5 × 10⁹ | 7.5 × 10⁹ | 2022 | architecture not stated | | 245 | OpenVLA (Stanford University, 2024) | 7.2 × 10⁹ | 7.2 × 10⁹ | 2024 | architecture not stated | | 246 | mPLUG-Owl2 (Alibaba, 2023) | \~7.1 × 10⁹ | | 2023 | estimate | | 247 | Goat-7B (National University of Singapore, 2023) | \~7 × 10⁹ | | 2023 | estimate | | 248 | Llama 2-7B (Meta AI, 2023) | 7 × 10⁹ | 7 × 10⁹ | 2023 | dense | | 249 | Llama Guard (Meta AI, 2023) | 7 × 10⁹ | 7 × 10⁹ | 2023 | dense | | 250 | Qwen-VL-Max (Alibaba, 2024) | 7 × 10⁹ | 7 × 10⁹ | 2024 | dense | | 251 | Mathstral (Mistral AI, 2024) | 7 × 10⁹ | 7 × 10⁹ | 2024 | architecture not stated | | 252 | Apollo 7B (Meta AI, 2024) | 7 × 10⁹ | 7 × 10⁹ | 2024 | architecture not stated | | 253 | Eurus-2-7B-PRIME (Tsinghua University, 2025) | \~7 × 10⁹ | | 2025 | estimate | | 254 | INTELLECT-MATH (Prime Intellect, 2025) | 7 × 10⁹ | 7 × 10⁹ | 2025 | architecture not stated | | 255 | Incoder-6.7B (Facebook AI Research, 2023) | 6.7 × 10⁹ | 6.7 × 10⁹ | 2023 | architecture not stated | | 256 | ProGen2-xlarge (Salesforce Research, 2022) | 6.4 × 10⁹ | 6.4 × 10⁹ | 2022 | architecture not stated | | 257 | GPipe (Google, 2018) | \~6 × 10⁹ | | 2018 | estimate | | 258 | InstructGPT 6B (OpenAI, 2022) | 6 × 10⁹ | 6 × 10⁹ | 2022 | dense | | 259 | Vega v2 (Wuhan University, 2022) | 6 × 10⁹ | 6 × 10⁹ | 2022 | architecture not stated | | 260 | ChatGLM3-6B (Z.ai (Zhipu AI), 2023) | \~6 × 10⁹ | | 2023 | estimate | | 261 | MolmoAct 2 (Allen Institute for AI, 2026) | \~5.5 × 10⁹ | | 2026 | estimate | | 262 | T-NLRv5 XXL (Microsoft, 2021) | 5.4 × 10⁹ | 5.4 × 10⁹ | 2021 | architecture not stated | | 263 | π0.6 (Physical Intelligence, 2025) | 5.3 × 10⁹ | 5.3 × 10⁹ | 2025 | mixture of experts (active count unpublished) | | 264 | PaLI-3 (Google DeepMind, 2023) | 5 × 10⁹ | 5 × 10⁹ | 2023 | architecture not stated | | 265 | Mesh-TensorFlow Transformer 4.9B (Google Brain, 2018) | 4.9 × 10⁹ | 4.9 × 10⁹ | 2018 | dense | | 266 | W.A.L.T (Stanford University, 2023) | 4.7 × 10⁹ | 4.7 × 10⁹ | 2023 | architecture not stated | | 267 | CogView (Tsinghua University, 2021) | \~4 × 10⁹ | | 2021 | estimate | | 268 | Make-A-Scene (Meta AI, 2022) | \~4 × 10⁹ | | 2022 | estimate | | 269 | ONE-PEACE (Alibaba, 2023) | \~4 × 10⁹ | | 2023 | estimate | | 270 | Megatron-BERT (Nvidia, 2019) | 3.9 × 10⁹ | 3.9 × 10⁹ | 2019 | dense | | 271 | DALL·E 2 (OpenAI, 2022) | \~3.5 × 10⁹ | | 2022 | estimate | | 272 | Stable Diffusion XL (Stability AI, 2023) | \~3.4 × 10⁹ | | 2023 | estimate | | 273 | MusicGen (Meta AI, 2023) | \~3.4 × 10⁹ | | 2023 | estimate | | 274 | Gemini Nano-2 (Google DeepMind, 2023) | 3.3 × 10⁹ | 3.3 × 10⁹ | 2023 | architecture not stated | | 275 | Zidong Taichu (Chinese Academy of Sciences, 2021) | 3.2 × 10⁹ | 3.2 × 10⁹ | 2021 | architecture not stated | | 276 | DeepNet (Microsoft Research, 2022) | 3.2 × 10⁹ | 3.2 × 10⁹ | 2022 | architecture not stated | | 277 | MS-ensemble-speech-recognition (Microsoft, 2016) | \~3.2 × 10⁹ | | 2016 | estimate | | 278 | BASIC-L (Google, 2021) | \~3.1 × 10⁹ | | 2021 | estimate | | 279 | BASIC-L + Lion (Google, 2023) | 3.1 × 10⁹ | 3.1 × 10⁹ | 2023 | architecture not stated | | 280 | LEP-AD (King Abdullah University of Science and Technology (KAUST), 2023) | 3 × 10⁹ | 3 × 10⁹ | 2023 | architecture not stated | | 281 | ViT-G/14 (Google Research, 2021) | 3 × 10⁹ | 3 × 10⁹ | 2021 | architecture not stated | | 282 | ProtT5-XL-U50 (Technical University of Munich, 2021) | 3 × 10⁹ | 3 × 10⁹ | 2021 | architecture not stated | | 283 | Swin Transformer V2 (Microsoft Research Asia, 2021) | 3 × 10⁹ | 3 × 10⁹ | 2021 | architecture not stated | | 284 | LongT5 (Google Research, 2021) | 3 × 10⁹ | 3 × 10⁹ | 2021 | architecture not stated | | 285 | ReALM (Apple, 2024) | 3 × 10⁹ | 3 × 10⁹ | 2024 | architecture not stated | | 286 | Mesh-TensorFlow Transformer 2.9B (Google Brain, 2018) | \~2.9 × 10⁹ | | 2018 | estimate | | 287 | T5-3B (Google, 2019) | 2.8 × 10⁹ | 2.8 × 10⁹ | 2019 | dense | | 288 | DPO on Pythia-2.8B (Stanford University, 2023) | 2.8 × 10⁹ | 2.8 × 10⁹ | 2023 | dense | | 289 | AFM-on-device (Apple, 2024) | 2.7 × 10⁹ | 2.7 × 10⁹ | 2024 | architecture not stated | | 290 | PolyCoder (Carnegie Mellon University (CMU), 2022) | \~2.7 × 10⁹ | | 2022 | estimate | | 291 | Hybrid H3-2.7B (Stanford University, 2022) | \~2.7 × 10⁹ | | 2022 | estimate | | 292 | GPT3-2.7B (Stanford University, 2023) | 2.7 × 10⁹ | 2.7 × 10⁹ | 2023 | architecture not stated | | 293 | Meena (Google Brain, 2020) | 2.6 × 10⁹ | 2.6 × 10⁹ | 2020 | dense | | 294 | CPM-Large (Tsinghua University, 2020) | 2.6 × 10⁹ | 2.6 × 10⁹ | 2020 | dense | | 295 | Nucleotide Transformer (Nvidia, 2023) | \~2.5 × 10⁹ | | 2023 | estimate | | 296 | Fugatto 1 (Nvidia, 2024) | 2.5 × 10⁹ | 2.5 × 10⁹ | 2024 | architecture not stated | | 297 | CoAtNet (Google, 2021) | 2.4 × 10⁹ | 2.4 × 10⁹ | 2021 | architecture not stated | | 298 | GShard (Google, 2020) | 2.3 × 10⁹ | 2.3 × 10⁹ | 2020 | dense | | 299 | SeamlessM4T (Facebook, 2023) | 2.3 × 10⁹ | 2.3 × 10⁹ | 2023 | architecture not stated | | 300 | Turing ULRv5 (Microsoft, 2021) | 2.2 × 10⁹ | 2.2 × 10⁹ | 2021 | architecture not stated | | 301 | CoCa (Google Research, 2022) | 2.1 × 10⁹ | 2.1 × 10⁹ | 2022 | architecture not stated | | 302 | MuseNet (OpenAI, 2019) | \~2 × 10⁹ | | 2019 | estimate | | 303 | Infinity (ByteDance, 2024) | 2 × 10⁹ | 2 × 10⁹ | 2024 | architecture not stated | | 304 | Seq2Seq LSTM (Google, 2014) | 1.9 × 10⁹ | 1.9 × 10⁹ | 2014 | dense | | 305 | BEIT-3 (Microsoft, 2022) | \~1.9 × 10⁹ | | 2022 | estimate | | 306 | Ankh\_large (Technical University of Munich, 2023) | 1.9 × 10⁹ | 1.9 × 10⁹ | 2023 | architecture not stated | | 307 | ViT-G/14 (Google Brain, 2021) | 1.8 × 10⁹ | 1.8 × 10⁹ | 2021 | architecture not stated | | 308 | ViT-G (University of Washington, 2022) | 1.8 × 10⁹ | 1.8 × 10⁹ | 2022 | architecture not stated | | 309 | Phenaki (University College London (UCL), 2022) | \~1.8 × 10⁹ | | 2022 | estimate | | 310 | Gemini Nano-1 (Google DeepMind, 2023) | 1.8 × 10⁹ | 1.8 × 10⁹ | 2023 | architecture not stated | | 311 | DistBelief Vision (Google, 2012) | \~1.7 × 10⁹ | | 2012 | estimate | | 312 | Whisper (OpenAI, 2022) | \~1.6 × 10⁹ | | 2022 | estimate | | 313 | GPT-2 (OpenAI, 2019) | \~1.5 × 10⁹ | | 2019 | estimate | | 314 | DeBERTa (Microsoft, 2021) | 1.5 × 10⁹ | 1.5 × 10⁹ | 2021 | architecture not stated | | 315 | AR-LDM (Alibaba, 2022) | 1.5 × 10⁹ | 1.5 × 10⁹ | 2022 | architecture not stated | | 316 | AudioLM (Google Research, 2023) | \~1.5 × 10⁹ | | 2023 | estimate | | 317 | BLIP-2 (Salesforce Research, 2023) | 1.5 × 10⁹ | 1.5 × 10⁹ | 2023 | architecture not stated | | 318 | LDM-1.45B (Heidelberg University, 2021) | 1.4 × 10⁹ | 1.4 × 10⁹ | 2021 | architecture not stated | | 319 | Stable Diffusion (Runway, 2022) | 1.4 × 10⁹ | 1.4 × 10⁹ | 2022 | dense | | 320 | SEER (Facebook AI Research, 2021) | 1.3 × 10⁹ | 1.3 × 10⁹ | 2021 | architecture not stated | | 321 | InstructGPT 1.3B (OpenAI, 2022) | 1.3 × 10⁹ | 1.3 × 10⁹ | 2022 | dense | | 322 | Megatron-LM (Nvidia, 2019) | 1.2 × 10⁹ | 1.2 × 10⁹ | 2019 | dense | | 323 | RDT-1B (Tsinghua University, 2024) | \~1.2 × 10⁹ | | 2024 | estimate | | 324 | Gato (DeepMind, 2022) | 1.2 × 10⁹ | 1.2 × 10⁹ | 2022 | architecture not stated | | 325 | DINOv2 (Facebook AI Research, 2023) | 1.1 × 10⁹ | 1.1 × 10⁹ | 2023 | architecture not stated | | 326 | Diffusion Renderer (Nvidia, 2025) | 1.1 × 10⁹ | 1.1 × 10⁹ | 2025 | architecture not stated | | 327 | InternImage (Shanghai AI Lab, 2022) | 1.1 × 10⁹ | 1.1 × 10⁹ | 2022 | architecture not stated | | 328 | EVA-01 (Beijing Academy of Artificial Intelligence / BAAI, 2022) | 1 × 10⁹ | 1 × 10⁹ | 2022 | architecture not stated | | 329 | Unsupervised High-level Feature Learner (Google, 2012) | \~1 × 10⁹ | | 2012 | estimate | | 330 | Conformer + Wav2vec 2.0 + Noisy Student (Google, 2020) | 1 × 10⁹ | 1 × 10⁹ | 2020 | dense | | 331 | HuBERT (Facebook AI Research, 2021) | \~1 × 10⁹ | | 2021 | estimate | | 332 | W2v-BERT (Google Brain, 2021) | 1 × 10⁹ | 1 × 10⁹ | 2021 | dense | | 333 | AudioGen (Meta AI, 2023) | \~1 × 10⁹ | | 2023 | estimate | | 334 | VideoMAE V2 (Nanjing University, 2023) | 1 × 10⁹ | 1 × 10⁹ | 2023 | architecture not stated | | 335 | TransE (Universite de Technologie de Compiègne – CNRS, 2013) | \~9.4 × 10⁸ | | 2013 | estimate | | 336 | ImageBind (Meta AI, 2023) | \~9.3 × 10⁸ | | 2023 | estimate | | 337 | Big Transfer (Google Brain, 2019) | \~9.3 × 10⁸ | | 2019 | estimate | | 338 | Florence (Microsoft, 2021) | 8.9 × 10⁸ | 8.9 × 10⁸ | 2021 | architecture not stated | | 339 | NÜWA (Microsoft Research, 2021) | 8.7 × 10⁸ | 8.7 × 10⁸ | 2021 | architecture not stated | | 340 | MusicLM (Google, 2023) | 8.6 × 10⁸ | 8.6 × 10⁸ | 2023 | architecture not stated | | 341 | ResNeXt-101 32x48d (Facebook, 2018) | 8.3 × 10⁸ | 8.3 × 10⁸ | 2018 | dense | | 342 | FixRes ResNeXt-101 WSL (Facebook AI, 2019) | \~8.3 × 10⁸ | | 2019 | estimate | | 343 | ALIGN (Google Research, 2021) | 8.2 × 10⁸ | 8.2 × 10⁸ | 2021 | architecture not stated | | 344 | Statement Curriculum Learning (OpenAI, 2022) | \~7.7 × 10⁸ | | 2022 | estimate | | 345 | CodeT5-large (Salesforce, 2022) | \~7.7 × 10⁸ | | 2022 | estimate | | 346 | DITTO (Tsinghua University, 2022) | 7.5 × 10⁸ | 7.5 × 10⁸ | 2022 | architecture not stated | | 347 | FGN (Google DeepMind, 2025) | 7.2 × 10⁸ | 7.2 × 10⁸ | 2025 | architecture not stated | | 348 | data2vec (Meta AI, 2022) | \~7.1 × 10⁸ | | 2022 | estimate | | 349 | Tranception (University of Oxford, 2022) | 7 × 10⁸ | 7 × 10⁸ | 2022 | architecture not stated | | 350 | Word2Vec (Google, 2013) | \~6.9 × 10⁸ | | 2013 | estimate | | 351 | DiT-XL/2 (New York University (NYU), 2023) | 6.8 × 10⁸ | 6.8 × 10⁸ | 2023 | architecture not stated | | 352 | DiT-XL/2 + CADS (ETH Zurich, 2023) | \~6.8 × 10⁸ | | 2023 | estimate | | 353 | OmegaPLM (Massachusetts Institute of Technology (MIT), 2022) | 6.7 × 10⁸ | 6.7 × 10⁸ | 2022 | architecture not stated | | 354 | XLM (Facebook, 2019) | \~6.7 × 10⁸ | | 2019 | estimate | | 355 | ESM1b (Facebook AI Research, 2020) | 6.5 × 10⁸ | 6.5 × 10⁸ | 2020 | dense | | 356 | Segment Anything Model (Meta AI, 2023) | 6.4 × 10⁸ | 6.4 × 10⁸ | 2023 | architecture not stated | | 357 | ViT-Huge/14 (Google Brain, 2020) | 6.3 × 10⁸ | 6.3 × 10⁸ | 2020 | dense | | 358 | Masked Autoencoders ViT-H (Facebook AI Research, 2021) | \~6.3 × 10⁸ | | 2021 | estimate | | 359 | Retrieval-Augmented Generator (Facebook, 2020) | 6.3 × 10⁸ | 6.3 × 10⁸ | 2020 | dense | | 360 | DLDL (University of Oxford, 2016) | \~5.6 × 10⁸ | | 2016 | estimate | | 361 | ADM (OpenAI, 2021) | 5.6 × 10⁸ | 5.6 × 10⁸ | 2021 | architecture not stated | | 362 | TrOCR (Beihang University, 2021) | 5.6 × 10⁸ | 5.6 × 10⁸ | 2021 | architecture not stated | | 363 | XLM-RoBERTa (Facebook AI, 2019) | 5.5 × 10⁸ | 5.5 × 10⁸ | 2019 | dense | | 364 | LUKE (University of Washington, 2020) | \~4.8 × 10⁸ | | 2020 | estimate | | 365 | EfficientNet-L2 (Google, 2019) | \~4.8 × 10⁸ | | 2019 | estimate | | 366 | Noisy Student (Carnegie Mellon University (CMU), 2019) | 4.8 × 10⁸ | 4.8 × 10⁸ | 2019 | dense | | 367 | Meta Pseudo Labels (Google Brain, 2021) | \~4.8 × 10⁸ | | 2021 | estimate | | 368 | EMDR (Mila - Quebec AI (originally Montreal Institute for Learning Algorithms), 2021) | 4.4 × 10⁸ | 4.4 × 10⁸ | 2021 | architecture not stated | | 369 | NMT Transformer 437M (Google, 2019) | 4.4 × 10⁸ | 4.4 × 10⁸ | 2019 | dense | | 370 | Adaptive Inputs + LayerDrop (Facebook AI Research, 2019) | 4.2 × 10⁸ | 4.2 × 10⁸ | 2019 | dense | | 371 | ProtBERT-BFD (Technical University of Munich, 2021) | 4.2 × 10⁸ | 4.2 × 10⁸ | 2021 | architecture not stated | | 372 | OntoProtein (Zhejiang University (ZJU), 2022) | \~4.2 × 10⁸ | | 2022 | estimate | | 373 | DeBERTaV3large + KEAR (Microsoft, 2022) | 4.2 × 10⁸ | 4.2 × 10⁸ | 2022 | architecture not stated | | 374 | BART-large (Facebook AI, 2019) | \~4.1 × 10⁸ | | 2019 | estimate | | 375 | SigLIP 400M (Google DeepMind, 2023) | 4 × 10⁸ | 4 × 10⁸ | 2023 | architecture not stated | | 376 | BERT-Large-CAS (Amazon, 2019) | \~4 × 10⁸ | | 2019 | estimate | | 377 | MoCo (Facebook AI, 2019) | \~3.8 × 10⁸ | | 2019 | estimate | | 378 | SimCLR (Google Brain, 2020) | \~3.8 × 10⁸ | | 2020 | estimate | | 379 | CLIP (OpenAI, 2021) | 3.7 × 10⁸ | 3.7 × 10⁸ | 2021 | architecture not stated | | 380 | RoBERTa Large (Facebook, 2019) | 3.5 × 10⁸ | 3.5 × 10⁸ | 2019 | dense | | 381 | AbLang (University of Oxford, 2022) | 3.5 × 10⁸ | 3.5 × 10⁸ | 2022 | architecture not stated | | 382 | GPT-2 Medium (Stanford University, 2022) | 3.5 × 10⁸ | 3.5 × 10⁸ | 2022 | dense | | 383 | VALL-E (Microsoft, 2023) | \~3.5 × 10⁸ | | 2023 | estimate | | 384 | BERT-Large (Google, 2018) | 3.4 × 10⁸ | 3.4 × 10⁸ | 2018 | dense | | 385 | XLNet (Carnegie Mellon University (CMU), 2019) | 3.4 × 10⁸ | 3.4 × 10⁸ | 2019 | dense | | 386 | ERNIE-GEN (Baidu, 2020) | \~3.4 × 10⁸ | | 2020 | estimate | | 387 | CamemBERT (Facebook, 2019) | 3.4 × 10⁸ | 3.4 × 10⁸ | 2019 | dense | | 388 | ELECTRA (Stanford University, 2020) | \~3.4 × 10⁸ | | 2020 | estimate | | 389 | German ELECTRA Large (deepset, 2020) | 3.4 × 10⁸ | 3.4 × 10⁸ | 2020 | dense | | 390 | ALM 1.0 (Beijing Academy of Artificial Intelligence / BAAI, 2022) | \~3.4 × 10⁸ | | 2022 | estimate | | 391 | MT-DNN (Microsoft, 2019) | 3.3 × 10⁸ | 3.3 × 10⁸ | 2019 | dense | | 392 | Fusion in Encoder (Samsung, 2022) | \~3.3 × 10⁸ | | 2022 | estimate | | 393 | wave2vec 2.0 LARGE (Facebook, 2020) | 3.2 × 10⁸ | 3.2 × 10⁸ | 2020 | dense | | 394 | CPC v2 (DeepMind, 2019) | \~3 × 10⁸ | | 2019 | estimate | | 395 | Transformer - LibriVox + Decoding/Rescoring (Facebook, 2019) | 3 × 10⁸ | 3 × 10⁸ | 2019 | dense | | 396 | DDPM-IP (Utrecht University, 2023) | \~3 × 10⁸ | | 2023 | estimate | | 397 | GNMT (Google, 2016) | \~2.8 × 10⁸ | | 2016 | estimate | | 398 | base LM+GNN+kNN (Shannon.AI, 2021) | 2.7 × 10⁸ | 2.7 × 10⁸ | 2021 | architecture not stated | | 399 | YOLO (University of Washington, 2015) | \~2.7 × 10⁸ | | 2015 | estimate | | 400 | Transformer-XL (Carnegie Mellon University (CMU), 2019) | \~2.6 × 10⁸ | | 2019 | estimate | | 401 | Transformer-XL + RMS dynamic eval (University of Edinburgh, 2019) | 2.6 × 10⁸ | 2.6 × 10⁸ | 2019 | dense | | 402 | Transformer-XL Large + Phrase Induction (Massachusetts Institute of Technology (MIT), 2019) | \~2.6 × 10⁸ | | 2019 | estimate | | 403 | Segatron-XL large, M=384 + HCP (Microsoft Research, 2022) | 2.6 × 10⁸ | 2.6 × 10⁸ | 2022 | architecture not stated | | 404 | Denoising Diffusion Probabilistic Models (University of California (UC) Berkeley, 2021) | 2.6 × 10⁸ | 2.6 × 10⁸ | 2021 | architecture not stated | | 405 | Pangu-Weather (Huawei, 2023) | 2.6 × 10⁸ | 2.6 × 10⁸ | 2023 | architecture not stated | | 406 | SPN (Google Brain, 2018) | 2.5 × 10⁸ | 2.5 × 10⁸ | 2018 | dense | | 407 | S4 (Stanford University, 2021) | \~2.5 × 10⁸ | | 2021 | estimate | | 408 | Base LM + kNN LM + Continuous Cache (Stanford University, 2019) | \~2.5 × 10⁸ | | 2019 | estimate | | 409 | ERNIE-Doc (Baidu, 2020) | 2.5 × 10⁸ | 2.5 × 10⁸ | 2020 | dense | | 410 | Adaptive Input Transformer + RD (Microsoft Research Asia, 2021) | \~2.5 × 10⁸ | | 2021 | estimate | | 411 | Transformer (Adaptive Input Embeddings) WT103 (Facebook AI Research, 2018) | 2.5 × 10⁸ | 2.5 × 10⁸ | 2018 | dense | | 412 | TaLK Convolution (Carleton University, 2020) | \~2.4 × 10⁸ | | 2020 | estimate | | 413 | ALBERT-xxlarge (Toyota Technological Institute at Chicago, 2020) | \~2.4 × 10⁸ | | 2020 | estimate | | 414 | SRU++ Large (ASAPP, 2021) | 2.3 × 10⁸ | 2.3 × 10⁸ | 2021 | architecture not stated | | 415 | GR-2 (ByteDance, 2024) | 2.3 × 10⁸ | 2.3 × 10⁸ | 2024 | architecture not stated | | 416 | MaskGIT (Google Research, 2022) | 2.3 × 10⁸ | 2.3 × 10⁸ | 2022 | architecture not stated | | 417 | MetNet (Google, 2020) | \~2.3 × 10⁸ | | 2020 | estimate | | 418 | FFN SwiGLU (Google, 2020) | 2.2 × 10⁸ | 2.2 × 10⁸ | 2020 | dense | | 419 | CodeT5-base (Salesforce, 2021) | \~2.2 × 10⁸ | | 2021 | estimate | | 420 | Transformer (Google Research, 2017) | 2.1 × 10⁸ | 2.1 × 10⁸ | 2017 | dense | | 421 | Sandwich Transformer (Allen Institute for AI, 2019) | \~2.1 × 10⁸ | | 2019 | estimate | | 422 | EfficientNetV2-XL (Google, 2021) | 2.1 × 10⁸ | 2.1 × 10⁸ | 2021 | architecture not stated | | 423 | Word2Vec (Google, 2013) | \~2.1 × 10⁸ | | 2013 | estimate | | 424 | Multilingual DNN (Google, 2013) | 2.1 × 10⁸ | 2.1 × 10⁸ | 2013 | dense | | 425 | DreamerV3 (DeepMind, 2023) | \~2 × 10⁸ | | 2023 | estimate | | 426 | DL scaling speech (Baidu, 2017) | \~1.9 × 10⁸ | | 2017 | estimate | | 427 | ResNeXt-101 Billion-scale (Facebook AI, 2019) | \~1.9 × 10⁸ | | 2019 | estimate | | 428 | (NTT Communication Science Laboratories, 2018) | 1.9 × 10⁸ | 1.9 × 10⁸ | 2018 | dense | | 429 | TrellisNet (Carnegie Mellon University (CMU), 2018) | 1.8 × 10⁸ | 1.8 × 10⁸ | 2018 | dense | | 430 | DL scaling LM (Baidu, 2017) | \~1.8 × 10⁸ | | 2017 | estimate | | 431 | EXAONE Path 2.0 (LG AI Research, 2025) | 1.8 × 10⁸ | 1.8 × 10⁸ | 2025 | architecture not stated | | 432 | OpenAI Five Rerun (OpenAI, 2019) | 1.6 × 10⁸ | 1.6 × 10⁸ | 2019 | dense | | 433 | OpenAI Five (OpenAI, 2019) | 1.6 × 10⁸ | 1.6 × 10⁸ | 2019 | dense | | 434 | 4 layer QRNN (Salesforce Research, 2018) | \~1.5 × 10⁸ | | 2018 | estimate | | 435 | TransformerXL + spectrum control (University of California Los Angeles (UCLA), 2020) | \~1.5 × 10⁸ | | 2020 | estimate | | 436 | AlphaGeometry (Google DeepMind, 2024) | 1.5 × 10⁸ | 1.5 × 10⁸ | 2024 | architecture not stated | | 437 | PermuteFormer (Peking University, 2021) | \~1.5 × 10⁸ | | 2021 | estimate | | 438 | Fragment embedding (Stanford University, 2014) | \~1.4 × 10⁸ | | 2014 | estimate | | 439 | OverFeat (New York University (NYU), 2013) | 1.4 × 10⁸ | 1.4 × 10⁸ | 2013 | dense | | 440 | VGG19 (University of Oxford, 2014) | \~1.4 × 10⁸ | | 2014 | estimate | | 441 | LRCN (UT Austin, 2014) | \~1.4 × 10⁸ | | 2014 | estimate | | 442 | Transformer-XL DeFINE (University of Washington, 2019) | \~1.4 × 10⁸ | | 2019 | estimate | | 443 | Deep Autoencoders (University of Toronto, 2011) | 1.4 × 10⁸ | 1.4 × 10⁸ | 2011 | dense | | 444 | AlphaStar (DeepMind, 2019) | 1.4 × 10⁸ | 1.4 × 10⁸ | 2019 | dense | | 445 | VGG16 (University of Oxford, 2014) | 1.4 × 10⁸ | 1.4 × 10⁸ | 2014 | dense | | 446 | VGG-Face (University of Oxford, 2015) | 1.4 × 10⁸ | 1.4 × 10⁸ | 2015 | dense | | 447 | SAF R-CNN (Beijing Institute of Technology, 2015) | \~1.4 × 10⁸ | | 2015 | estimate | | 448 | Template Adaptation (University of Oxford, 2016) | \~1.4 × 10⁸ | | 2016 | estimate | | 449 | LRR-4X (UC Irvine, 2016) | \~1.4 × 10⁸ | | 2016 | estimate | | 450 | CMS-RCNN (IEEE, 2016) | \~1.4 × 10⁸ | | 2016 | estimate | | 451 | QRNN (Salesforce Research, 2018) | \~1.4 × 10⁸ | | 2018 | estimate | | 452 | Multiresolution CNN (Google, 2014) | \~1.3 × 10⁸ | | 2014 | estimate | | 453 | FTW (DeepMind, 2018) | \~1.3 × 10⁸ | | 2018 | estimate | | 454 | Feedback Transformer (LORIA, 2020) | \~1.3 × 10⁸ | | 2020 | estimate | | 455 | KEPLER (Tsinghua University, 2020) | 1.3 × 10⁸ | 1.3 × 10⁸ | 2020 | dense | | 456 | DL scaling Image (Baidu, 2017) | 1.2 × 10⁸ | 1.2 × 10⁸ | 2017 | dense | | 457 | GloVe (Stanford University, 2014) | \~1.2 × 10⁸ | | 2014 | estimate | | 458 | Conformer (Google, 2020) | 1.2 × 10⁸ | 1.2 × 10⁸ | 2020 | dense | | 459 | GPT-1 (OpenAI, 2018) | \~1.2 × 10⁸ | | 2018 | estimate | | 460 | ContextNet (Google, 2020) | \~1.1 × 10⁸ | | 2020 | estimate | | 461 | SciBERT (Allen Institute for AI, 2019) | 1.1 × 10⁸ | 1.1 × 10⁸ | 2019 | dense | | 462 | DNABERT (Northeastern University, 2021) | 1.1 × 10⁸ | 1.1 × 10⁸ | 2021 | architecture not stated | | 463 | Contriever (Meta AI, 2021) | \~1.1 × 10⁸ | | 2021 | estimate | | 464 | BERT-RBP (Waseda University, 2022) | 1.1 × 10⁸ | 1.1 × 10⁸ | 2022 | dense | | 465 | RoFormer (Zhuiyi Technology, 2023) | 1.1 × 10⁸ | 1.1 × 10⁸ | 2023 | architecture not stated | | 466 | LSTM LM (RWTH Aachen University, 2012) | \~1 × 10⁸ | | 2012 | estimate | | 467 | ReLU-Speech (Google, 2013) | \~1 × 10⁸ | | 2013 | estimate | | 468 | STORM-B/8 (University of Southern California, 2024) | 1 × 10⁸ | 1 × 10⁸ | 2024 | architecture not stated | | 469 | Support Vector Machines (AT&T, 1995) | \~1 × 10⁸ | | 1995 | estimate | | 470 | GPU DBNs (Stanford University, 2009) | 1 × 10⁸ | 1 × 10⁸ | 2009 | dense | | 471 | MSA Transformer (Facebook AI Research, 2021) | \~1 × 10⁸ | | 2021 | estimate | | 472 | YOLOX-X (Megvii Inc, 2021) | \~9.9 × 10⁷ | | 2021 | estimate | | 473 | DeLighT (University of Washington, 2020) | \~9.9 × 10⁷ | | 2020 | estimate | | 474 | ELMo (University of Washington, 2018) | \~9.4 × 10⁷ | | 2018 | estimate | | 475 | AlphaFold 2 (DeepMind, 2020) | \~9.3 × 10⁷ | | 2020 | estimate | | 476 | AlphaMissense (Google DeepMind, 2023) | \~9.3 × 10⁷ | | 2023 | estimate | | 477 | Octo-Base (University of California (UC) Berkeley, 2024) | 9.3 × 10⁷ | 9.3 × 10⁷ | 2024 | architecture not stated | | 478 | PolyNet (Chinese University of Hong Kong (CUHK), 2016) | \~9.2 × 10⁷ | | 2016 | estimate | | 479 | Transformer local-attention (Google Cloud, 2021) | 9 × 10⁷ | 9 × 10⁷ | 2021 | architecture not stated | | 480 | Transformer + Simple Recurrent Unit (ASAPP, 2018) | 9 × 10⁷ | 9 × 10⁷ | 2018 | dense | | 481 | NASNet-A (Google Brain, 2017) | \~8.9 × 10⁷ | | 2017 | estimate | | 482 | CLIP (OpenAI, 2021) | \~8.9 × 10⁷ | | 2021 | estimate | | 483 | Detic (Meta AI, 2022) | \~8.8 × 10⁷ | | 2022 | estimate | | 484 | MSRA (Microsoft Research, 2015) | 8.7 × 10⁷ | 8.7 × 10⁷ | 2015 | dense | | 485 | BigBiGAN (Google, 2019) | \~8.6 × 10⁷ | | 2019 | estimate | | 486 | ViT-Base/32 (Google Brain, 2020) | \~8.6 × 10⁷ | | 2020 | estimate | | 487 | DeiT-B (Meta AI, 2021) | 8.6 × 10⁷ | 8.6 × 10⁷ | 2021 | architecture not stated | | 488 | CaLM (University of Oxford, 2022) | \~8.6 × 10⁷ | | 2022 | estimate | | 489 | Tensor-Transformer (University of California (UC) Berkeley, 2020) | 8.5 × 10⁷ | 8.5 × 10⁷ | 2020 | dense | | 490 | ResNeXt-101 (University of California San Diego, 2016) | 8.3 × 10⁷ | 8.3 × 10⁷ | 2016 | dense | | 491 | RBM Image Classifier (University of Toronto, 2009) | \~8 × 10⁷ | | 2009 | estimate | | 492 | Routing Transformer (Google Research, 2020) | 8 × 10⁷ | 8 × 10⁷ | 2020 | dense | | 493 | Big-Little Net (IBM, 2018) | \~7.7 × 10⁷ | | 2018 | estimate | | 494 | EfficientDet (Google Brain, 2020) | \~7.7 × 10⁷ | | 2020 | estimate | | 495 | CODEFUSION (Microsoft, 2023) | 7.5 × 10⁷ | 7.5 × 10⁷ | 2023 | architecture not stated | | 496 | RNN LM (Johns Hopkins University, 2010) | \~7 × 10⁷ | | 2010 | estimate | | 497 | R-CNN (University of California (UC) Berkeley, 2013) | \~6.9 × 10⁷ | | 2013 | estimate | | 498 | Variational (University of Cambridge, 2015) | 6.6 × 10⁷ | 6.6 × 10⁷ | 2015 | dense | | 499 | DistilBERT (Hugging Face, 2019) | 6.6 × 10⁷ | 6.6 × 10⁷ | 2019 | dense | | 500 | Theseus 6/768 (University of California San Diego, 2020) | 6.6 × 10⁷ | 6.6 × 10⁷ | 2020 | dense | | 501 | ResNet-152 (Microsoft, 2015) | 6 × 10⁷ | 6 × 10⁷ | 2015 | dense | | 502 | AlexNet (University of Toronto, 2012) | 6 × 10⁷ | 6 × 10⁷ | 2012 | dense | | 503 | DETR (Facebook, 2020) | 6 × 10⁷ | 6 × 10⁷ | 2020 | dense | | 504 | YOLOv3 (University of Washington, 2018) | \~5.7 × 10⁷ | | 2018 | estimate | | 505 | Inception-ResNet-V2 (Google, 2016) | \~5.6 × 10⁷ | | 2016 | estimate | | 506 | Transformer ELMo (Allen Institute for AI, 2019) | 5.6 × 10⁷ | 5.6 × 10⁷ | 2019 | dense | | 507 | Fruit fly brain | 5.5 × 10⁷ | 5.5 × 10⁷ | | synapses | | 508 | NAS with base 8 and shared embeddings (Google Brain, 2016) | 5.4 × 10⁷ | 5.4 × 10⁷ | 2016 | dense | | 509 | RetinaNet-R101 (Facebook AI Research, 2017) | 5.3 × 10⁷ | 5.3 × 10⁷ | 2017 | dense | | 510 | AWD-LSTM+WT+Cache+IOG (NTT Communication Science Laboratories, 2017) | 5.3 × 10⁷ | 5.3 × 10⁷ | 2017 | dense | | 511 | Hierarchical Scene Labeling (New York University (NYU), 2013) | 5.2 × 10⁷ | 5.2 × 10⁷ | 2013 | dense | | 512 | VD-LSTM+REAL Large (Salesforce Research, 2016) | 5.1 × 10⁷ | 5.1 × 10⁷ | 2016 | dense | | 513 | YOLOv2 (University of Washington, 2016) | \~5.1 × 10⁷ | | 2016 | estimate | | 514 | Dropout-LSTM+Noise (Columbia University, 2018) | \~5.1 × 10⁷ | | 2018 | estimate | | 515 | LSTM + dynamic eval (University of Edinburgh, 2017) | \~5 × 10⁷ | | 2017 | estimate | | 516 | Dropout (University of Toronto, 2012) | \~4.9 × 10⁷ | | 2012 | estimate | | 517 | Dropout: SVHN (University of Toronto, 2014) | 4.8 × 10⁷ | 4.8 × 10⁷ | 2014 | dense | | 518 | DistBelief Speech (Google, 2012) | \~4.7 × 10⁷ | | 2012 | estimate | | 519 | AlphaGo Zero (DeepMind, 2017) | 4.6 × 10⁷ | 4.6 × 10⁷ | 2017 | dense | | 520 | Fine-tuned-AWD-LSTM-DOC (Samsung R&D Institute Russia, 2018) | 4.6 × 10⁷ | 4.6 × 10⁷ | 2018 | dense | | 521 | CT-MoS (Google, 2020) | 4.5 × 10⁷ | 4.5 × 10⁷ | 2020 | dense | | 522 | JFT (Google Research, 2017) | 4.5 × 10⁷ | 4.5 × 10⁷ | 2017 | dense | | 523 | ResNet-101 (Microsoft, 2015) | 4.5 × 10⁷ | 4.5 × 10⁷ | 2015 | dense | | 524 | HR-ResNet101 (Carnegie Mellon University (CMU), 2016) | 4.5 × 10⁷ | 4.5 × 10⁷ | 2016 | dense | | 525 | 3DMM-CNN (University of Southern California, 2016) | 4.5 × 10⁷ | 4.5 × 10⁷ | 2016 | dense | | 526 | LaNet-L (Brown University, 2019) | 4.4 × 10⁷ | 4.4 × 10⁷ | 2019 | dense | | 527 | ProBERTa (University of Illinois Urbana-Champaign (UIUC), 2020) | 4.4 × 10⁷ | 4.4 × 10⁷ | 2020 | dense | | 528 | PreTrans-3L-250H (University of Toronto, 2013) | \~4.3 × 10⁷ | | 2013 | estimate | | 529 | Inceptionv4 (Google, 2016) | \~4.3 × 10⁷ | | 2016 | estimate | | 530 | GL-LWGC-AWD-MoS-LSTM + dynamic evaluation (Ben-Gurion University of the Negev, 2017) | 3.8 × 10⁷ | 3.8 × 10⁷ | 2017 | dense | | 531 | U-Net (University of Freiburg, 2015) | 3.8 × 10⁷ | 3.8 × 10⁷ | 2015 | dense | | 532 | NASv3 (Google Brain, 2016) | \~3.7 × 10⁷ | | 2016 | estimate | | 533 | MuZero (DeepMind, 2019) | \~3.7 × 10⁷ | | 2019 | estimate | | 534 | AWD-LSTM-MoS + dynamic evaluation (Carnegie Mellon University (CMU), 2017) | \~3.5 × 10⁷ | | 2017 | estimate | | 535 | AWD-LSTM-MoS+PDR + dynamic evaluation (IBM, 2018) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2018 | dense | | 536 | AWD-LSTM-MoS + dynamic evaluation (Peking University, 2018) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2018 | dense | | 537 | AWD-LSTM + MoS + Partial Shuffled (University of Texas at Austin, 2019) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2019 | dense | | 538 | Mogrifier (d2, MoS2, MC) + dynamic eval (DeepMind, 2019) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2019 | dense | | 539 | Mogrifier RLSTM (DeepMind, 2022) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2022 | architecture not stated | | 540 | RT-1 (Google, 2022) | 3.5 × 10⁷ | 3.5 × 10⁷ | 2022 | architecture not stated | | 541 | DenseNet-264 (Tsinghua University, 2016) | 3.4 × 10⁷ | 3.4 × 10⁷ | 2016 | dense | | 542 | RetinaNet-R50 (Facebook AI Research, 2017) | \~3.4 × 10⁷ | | 2017 | estimate | | 543 | Fraternal dropout + AWD-LSTM 3-layer (Jagiellonian University, 2017) | \~3.4 × 10⁷ | | 2017 | estimate | | 544 | AWD-LSTM-DRILL + dynamic evaluation† (IDIAP, 2019) | \~3.4 × 10⁷ | | 2019 | estimate | | 545 | AWD-LSTM - 3-layer LSTM (Salesforce Research, 2017) | 3.3 × 10⁷ | 3.3 × 10⁷ | 2017 | dense | | 546 | TCAN (Nanjing University, 2020) | 3.3 × 10⁷ | 3.3 × 10⁷ | 2020 | dense | | 547 | MMLSTM (Beijing University of Posts and Telecommunications, 2019) | \~3.2 × 10⁷ | | 2019 | estimate | | 548 | Paragraph Vector (Google, 2014) | 3.2 × 10⁷ | 3.2 × 10⁷ | 2014 | dense | | 549 | aLSTM (University of Manchester, 2018) | 3.2 × 10⁷ | 3.2 × 10⁷ | 2018 | dense | | 550 | StyleGAN2 (Nvidia, 2019) | 3 × 10⁷ | 3 × 10⁷ | 2019 | dense | | 551 | UDSMProt (Fraunhofer Heinrich Hertz Institute, 2019) | \~2.8 × 10⁷ | | 2019 | estimate | | 552 | SENet (Chinese Academy of Sciences, 2017) | \~2.8 × 10⁷ | | 2017 | estimate | | 553 | Perceiver IO (DeepMind, 2020) | 2.8 × 10⁷ | 2.8 × 10⁷ | 2020 | dense | | 554 | Fractional Max-Pooling (University of Warwick, 2014) | \~2.7 × 10⁷ | | 2014 | estimate | | 555 | Hierarchical Reasoning Model (Sapient Intelligence, 2025) | 2.7 × 10⁷ | 2.7 × 10⁷ | 2025 | architecture not stated | | 556 | RNN-WER (DeepMind, 2014) | \~2.6 × 10⁷ | | 2014 | estimate | | 557 | StyleGAN (Nvidia, 2018) | 2.6 × 10⁷ | 2.6 × 10⁷ | 2018 | dense | | 558 | PyramidNet (Korea Advanced Institute of Science and Technology (KAIST), 2017) | \~2.6 × 10⁷ | | 2017 | estimate | | 559 | ResNeXt-50 (University of California San Diego, 2016) | \~2.5 × 10⁷ | | 2016 | estimate | | 560 | AWD-LSTM (DeepMind, 2017) | 2.4 × 10⁷ | 2.4 × 10⁷ | 2017 | dense | | 561 | Inception v3 (Google, 2015) | \~2.4 × 10⁷ | | 2015 | estimate | | 562 | Mamba-24M (Carnegie Mellon University (CMU), 2023) | 2.3 × 10⁷ | 2.3 × 10⁷ | 2023 | architecture not stated | | 563 | WeNet (Amazon, 2019) | \~2.3 × 10⁷ | | 2019 | estimate | | 564 | EN^2AS with performance reward (Beijing Institute of Technology, 2019) | \~2.3 × 10⁷ | | 2019 | estimate | | 565 | NAS+ESS (Northeastern University (China), 2020) | 2.3 × 10⁷ | 2.3 × 10⁷ | 2020 | dense | | 566 | Xception (Google, 2016) | 2.3 × 10⁷ | 2.3 × 10⁷ | 2016 | dense | | 567 | MetaMimic (Google, 2018) | \~2.2 × 10⁷ | | 2018 | estimate | | 568 | MMLSTM (Beijing University of Posts and Telecommunications, 2019) | \~2.1 × 10⁷ | | 2019 | estimate | | 569 | Pointer Sentinel-LSTM (MetaMind Inc, 2016) | 2.1 × 10⁷ | 2.1 × 10⁷ | 2016 | dense | | 570 | DiffDock (Massachusetts Institute of Technology (MIT), 2022) | \~2 × 10⁷ | | 2022 | estimate | | 571 | EI-REHN-1000D (Korea Advanced Institute of Science and Technology (KAIST), 2017) | 1.9 × 10⁷ | 1.9 × 10⁷ | 2017 | dense | | 572 | TC-DNN-BLSTM-DNN (Carnegie Mellon University (CMU), 2015) | 1.8 × 10⁷ | 1.8 × 10⁷ | 2015 | dense | | 573 | OR-WideResNet (Duke University, 2017) | 1.8 × 10⁷ | 1.8 × 10⁷ | 2017 | dense | | 574 | BP-DBN (University of Toronto, 2009) | 1.8 × 10⁷ | 1.8 × 10⁷ | 2009 | dense | | 575 | ALBERT (Toyota Technological Institute at Chicago, 2019) | 1.8 × 10⁷ | 1.8 × 10⁷ | 2019 | dense | | 576 | MedBERT (Peng Cheng Laboratory, 2021) | \~1.7 × 10⁷ | | 2021 | estimate | | 577 | LCNP NORB (, 2009) | 1.7 × 10⁷ | 1.7 × 10⁷ | 2009 | dense | | 578 | LF-MMI (Johns Hopkins University, 2016) | 1.7 × 10⁷ | 1.7 × 10⁷ | 2016 | dense | | 579 | AlphaFold (DeepMind, 2020) | \~1.6 × 10⁷ | | 2020 | estimate | | 580 | GNoME for crystal discovery (Google DeepMind, 2023) | \~1.6 × 10⁷ | | 2023 | estimate | | 581 | ReLU (University of Toronto, 2010) | \~1.6 × 10⁷ | | 2010 | estimate | | 582 | ProteinBERT (Hebrew University of Jerusalem, 2022) | 1.6 × 10⁷ | 1.6 × 10⁷ | 2022 | architecture not stated | | 583 | Eve (Harvard Medical School, 2021) | \~1.5 × 10⁷ | | 2021 | estimate | | 584 | LCNP LabelMe (University of Bonn, 2009) | 1.4 × 10⁷ | 1.4 × 10⁷ | 2009 | dense | | 585 | BatchNorm (Google, 2015) | 1.4 × 10⁷ | 1.4 × 10⁷ | 2015 | dense | | 586 | Fold2Seq (IBM, 2021) | \~1.2 × 10⁷ | | 2021 | estimate | | 587 | NPLM (University of Montreal / Université de Montréal, 2003) | 1.2 × 10⁷ | 1.2 × 10⁷ | 2003 | dense | | 588 | LCNP MNIST (, 2009) | 1.2 × 10⁷ | 1.2 × 10⁷ | 2009 | dense | | 589 | ISS (Duke University, 2017) | 1.1 × 10⁷ | 1.1 × 10⁷ | 2017 | dense | | 590 | ResNet-1001 (Microsoft, 2016) | \~1 × 10⁷ | | 2016 | estimate | | 591 | RNN+LDA+KN5+cache (Microsoft, 2012) | \~9 × 10⁶ | | 2012 | estimate | | 592 | Gated HORNN (York University, 2016) | \~9 × 10⁶ | | 2016 | estimate | | 593 | AlphaGo Fan (DeepMind, 2015) | \~8.2 × 10⁶ | | 2015 | estimate | | 594 | genCNN + dyn eval (Chinese Academy of Sciences, 2015) | \~8 × 10⁶ | | 2015 | estimate | | 595 | Once for All (MIT-IBM Watson AI Lab, 2020) | 7.7 × 10⁶ | 7.7 × 10⁶ | 2020 | dense | | 596 | Multi-cell LSTM (University of Hyderabad, 2018) | \~7.2 × 10⁶ | | 2018 | estimate | | 597 | Feedforward NN (University of Montreal / Université de Montréal, 2010) | 7.1 × 10⁶ | 7.1 × 10⁶ | 2010 | dense | | 598 | True-Regularization+Finetune+Dynamic-Eval (Mobvoi, 2019) | 7 × 10⁶ | 7 × 10⁶ | 2019 | dense | | 599 | Neural LM (University of Montreal / Université de Montréal, 2000) | 6.9 × 10⁶ | 6.9 × 10⁶ | 2000 | dense | | 600 | GoogLeNet / InceptionV1 (Google, 2014) | 6.8 × 10⁶ | 6.8 × 10⁶ | 2014 | dense | | 601 | HyenaDNA (Stanford University, 2023) | 6.6 × 10⁶ | 6.6 × 10⁶ | 2023 | architecture not stated | | 602 | Dropout (University of Toronto, 2012) | 5.6 × 10⁶ | 5.6 × 10⁶ | 2012 | dense | | 603 | SimpleNet (Sensifai, 2016) | 5.5 × 10⁶ | 5.5 × 10⁶ | 2016 | dense | | 604 | AlphaX-1 (Facebook AI Research, 2019) | 5.4 × 10⁶ | 5.4 × 10⁶ | 2019 | dense | | 605 | 3DDFA (Chinese Academy of Sciences, 2015) | 5.4 × 10⁶ | 5.4 × 10⁶ | 2015 | dense | | 606 | Deep CNN + COTS (IEEE, 2015) | 5 × 10⁶ | 5 × 10⁶ | 2015 | dense | | 607 | DCNN (University of Maryland, 2015) | 5 × 10⁶ | 5 × 10⁶ | 2015 | dense | | 608 | NLP from scratch (NEC Laboratories, 2011) | \~5 × 10⁶ | | 2011 | estimate | | 609 | SPN-4+KN5 (Singapore University of Technology & Design, 2014) | \~5 × 10⁶ | | 2014 | estimate | | 610 | 6-Act Tether (Facebook AI Research, 2021) | 5 × 10⁶ | 5 × 10⁶ | 2021 | architecture not stated | | 611 | High Performance CNN (IDSIA, 2011) | \~4.9 × 10⁶ | | 2011 | estimate | | 612 | MobileNet (Google, 2017) | \~4.2 × 10⁶ | | 2017 | estimate | | 613 | NPLM (University of Montreal / Université de Montréal, 2003) | 4.1 × 10⁶ | 4.1 × 10⁶ | 2003 | dense | | 614 | LMICA (, 2004) | 4.1 × 10⁶ | 4.1 × 10⁶ | 2004 | dense | | 615 | PoE MNIST (University College London (UCL), 2000) | 3.9 × 10⁶ | 3.9 × 10⁶ | 2000 | dense | | 616 | Dimensionality Reduction (University of Toronto, 2006) | \~3.8 × 10⁶ | | 2006 | estimate | | 617 | GOAT (DeepMind, 2021) | \~3.5 × 10⁶ | | 2021 | estimate | | 618 | MobileNetV2 (Google, 2018) | \~3.4 × 10⁶ | | 2018 | estimate | | 619 | Big-Little Net (IBM, 2018) | \~3.3 × 10⁶ | | 2018 | estimate | | 620 | DLWP (University of Washington, 2021) | 2.7 × 10⁶ | 2.7 × 10⁶ | 2021 | architecture not stated | | 621 | BIDAF (University of Washington, 2016) | 2.6 × 10⁶ | 2.6 × 10⁶ | 2016 | dense | | 622 | DeepStack (University of Alberta, 2017) | \~2.5 × 10⁶ | | 2017 | estimate | | 623 | KataGo (Jane Street, 2019) | \~2.5 × 10⁶ | | 2019 | estimate | | 624 | ShuffleNet v1 (Megvii Inc, 2017) | \~2.4 × 10⁶ | | 2017 | estimate | | 625 | ADAM (University of Amsterdam, 2014) | 2.4 × 10⁶ | 2.4 × 10⁶ | 2014 | dense | | 626 | StyleGAN3-T (Nvidia, 2021) | 2.2 × 10⁶ | 2.2 × 10⁶ | 2021 | architecture not stated | | 627 | LSTM+NeuralCache (KU Leuven, 2018) | \~2.1 × 10⁶ | | 2018 | estimate | | 628 | HLBL (University of Toronto, 2008) | 1.8 × 10⁶ | 1.8 × 10⁶ | 2008 | dense | | 629 | ResNet-110 (Microsoft, 2015) | \~1.7 × 10⁶ | | 2015 | estimate | | 630 | Dueling DQN (Google DeepMind, 2016) | 1.7 × 10⁶ | 1.7 × 10⁶ | 2016 | dense | | 631 | DQN-2015 (Google, 2015) | \~1.7 × 10⁶ | | 2015 | estimate | | 632 | Rational DQN Average (TU Darmstadt, 2021) | \~1.7 × 10⁶ | | 2021 | estimate | | 633 | IBM-5 (IBM, 1993) | \~1.7 × 10⁶ | | 1993 | estimate | | 634 | Deep Belief Nets (University of Toronto, 2006) | \~1.6 × 10⁶ | | 2006 | estimate | | 635 | IMPALA (DeepMind, 2018) | 1.6 × 10⁶ | 1.6 × 10⁶ | 2018 | dense | | 636 | StyleGAN3-R (Nvidia, 2021) | 1.6 × 10⁶ | 1.6 × 10⁶ | 2021 | architecture not stated | | 637 | Deep Multitask NLP Network (NEC Laboratories, 2008) | \~1.5 × 10⁶ | | 2008 | estimate | | 638 | Double DQN (Google DeepMind, 2016) | 1.5 × 10⁶ | 1.5 × 10⁶ | 2016 | dense | | 639 | CNN committee (IDSIA, 2011) | \~1.4 × 10⁶ | | 2011 | estimate | | 640 | SqueezeNet (DeepScale, 2016) | \~1.2 × 10⁶ | | 2016 | estimate | | 641 | Neocognitron (NHK Broadcasting Science Research Laboratories, 1980) | 1.1 × 10⁶ | 1.1 × 10⁶ | 1980 | dense | | 642 | CURL (University of California (UC) Berkeley, 2020) | \~9.1 × 10⁵ | | 2020 | estimate | | 643 | DQN (DeepMind, 2013) | 8.4 × 10⁵ | 8.4 × 10⁵ | 2013 | dense | | 644 | EnhanceNet (Max Planck Institute for Intelligent Systems, 2016) | 8.1 × 10⁵ | 8.1 × 10⁵ | 2016 | dense | | 645 | Image generation (University of Amsterdam, 2013) | 7.8 × 10⁵ | 7.8 × 10⁵ | 2013 | dense | | 646 | Hanabi 4 player (DeepMind, 2019) | 7.6 × 10⁵ | 7.6 × 10⁵ | 2019 | dense | | 647 | TA-CNN (Chinese University of Hong Kong (CUHK), 2014) | \~7.1 × 10⁵ | | 2014 | estimate | | 648 | Robot Parkour (Shanghai Qi Zhi institute, 2023) | 5 × 10⁵ | 5 × 10⁵ | 2023 | architecture not stated | | 649 | LISSOM (University of Texas at Austin, 1995) | \~4.3 × 10⁵ | | 1995 | estimate | | 650 | SPIDER2 (Griffith University, 2016) | \~4.1 × 10⁵ | | 2016 | estimate | | 651 | Mixture of linear models (, 1994) | \~3.8 × 10⁵ | | 1994 | estimate | | 652 | NPD (IEEE, 2014) | 3.1 × 10⁵ | 3.1 × 10⁵ | 2014 | dense | | 653 | Pooling CNN (University of Bonn, 2010) | 2.9 × 10⁵ | 2.9 × 10⁵ | 2010 | dense | | 654 | Pooling CNN (University of Bonn, 2010) | 2.7 × 10⁵ | 2.7 × 10⁵ | 2010 | dense | | 655 | Two Stage Feature Extraction (New York University (NYU), 2009) | 2.6 × 10⁵ | 2.6 × 10⁵ | 2009 | dense | | 656 | Vector Space Model (Stanford University, 2011) | 2.5 × 10⁵ | 2.5 × 10⁵ | 2011 | dense | | 657 | HyperNEAT (University of Texas at Austin, 2014) | \~2.4 × 10⁵ | | 2014 | estimate | | 658 | DNN EM segmentation (IDSIA, 2012) | 2.2 × 10⁵ | 2.2 × 10⁵ | 2012 | dense | | 659 | Predictive Coding NN (Technical University of Munich, 1994) | 2.1 × 10⁵ | 2.1 × 10⁵ | 1994 | dense | | 660 | Table Tennis Agent (Google DeepMind, 2024) | \~1.9 × 10⁵ | | 2024 | estimate | | 661 | CNN Committee (IDSIA, 2011) | \~1.3 × 10⁵ | | 2011 | estimate | | 662 | CNN Committee (IDSIA, 2011) | \~1.2 × 10⁵ | | 2011 | estimate | | 663 | RECONTRA-uncategorized (, 1999) | \~1.1 × 10⁵ | | 1999 | estimate | | 664 | BLSTM for handwriting (University of Bern, 2007) | \~1 × 10⁵ | | 2007 | estimate | | 665 | SVM-CNN (New York University (NYU), 2006) | 9.1 × 10⁴ | 9.1 × 10⁴ | 2006 | dense | | 666 | Invariant CNN (New York University (NYU), 2004) | 9.1 × 10⁴ | 9.1 × 10⁴ | 2004 | dense | | 667 | SexNet compression (, 1990) | 7.3 × 10⁴ | 7.3 × 10⁴ | 1990 | dense | | 668 | NeuroChess (, 1994) | \~7.2 × 10⁴ | | 1994 | estimate | | 669 | RECONTRA-categorized (, 1999) | \~6.7 × 10⁴ | | 1999 | estimate | | 670 | LeNet-5 (AT&T, 1998) | 6 × 10⁴ | 6 × 10⁴ | 1998 | dense | | 671 | Swift (Intel Labs, 2023) | \~5.7 × 10⁴ | | 2023 | estimate | | 672 | TCN (Carnegie Mellon University (CMU), 2018) | 4.2 × 10⁴ | 4.2 × 10⁴ | 2018 | dense | | 673 | TRPO (University of California (UC) Berkeley, 2015) | 3.4 × 10⁴ | 3.4 × 10⁴ | 2015 | dense | | 674 | SOM-CNN (, 1997) | 3.2 × 10⁴ | 3.2 × 10⁴ | 1997 | dense | | 675 | NETtalk reimplementation (Oregon State University, 1990) | 2.7 × 10⁴ | 2.7 × 10⁴ | 1990 | dense | | 676 | TD-Gammon (IBM, 1992) | \~2.5 × 10⁴ | | 1992 | estimate | | 677 | Cognitron (Biological Cybernetics, 1975) | 2.2 × 10⁴ | 2.2 × 10⁴ | 1975 | dense | | 678 | NetTalk (Princeton University, 1987) | 1.9 × 10⁴ | 1.9 × 10⁴ | 1987 | dense | | 679 | CFSS (SenseTime, 2015) | \~1.7 × 10⁴ | | 2015 | estimate | | 680 | MUSIC perceptron (, 1996) | 1.4 × 10⁴ | 1.4 × 10⁴ | 1996 | dense | | 681 | AdaRNN (Beihang University, 2014) | 1.3 × 10⁴ | 1.3 × 10⁴ | 2014 | dense | | 682 | Bidirectional RNN (Advanced Telecommunications Research Institute, 1997) | \~1.3 × 10⁴ | | 1997 | estimate | | 683 | Decision tree (Mitsubishi Electric Research Labs, 2001) | \~1.2 × 10⁴ | | 2001 | estimate | | 684 | LSTM (Technical University of Munich, 1997) | 1.1 × 10⁴ | 1.1 × 10⁴ | 1997 | dense | | 685 | MLN-ASR (McGill University, 1988) | 1 × 10⁴ | 1 × 10⁴ | 1988 | dense | | 686 | Hopfield network (California Institute of Technology, 1982) | \~9.9 × 10³ | | 1982 | estimate | | 687 | Zip CNN (AT&T, 1989) | 9.8 × 10³ | 9.8 × 10³ | 1989 | dense | | 688 | Hierarchical Cognitron (NHK Broadcasting Science Research Laboratories, 1984) | \~9.3 × 10³ | | 1984 | estimate | | 689 | Weight Decay (, 1991) | 8.4 × 10³ | 8.4 × 10³ | 1991 | dense | | 690 | Deep Blue (IBM, 1997) | \~8 × 10³ | | 1997 | estimate | | 691 | Nematode (C. elegans) | 7 × 10³ | 7 × 10³ | | synapses | | 692 | System 11 (Carnegie Mellon University (CMU), 1996) | 6.5 × 10³ | 6.5 × 10³ | 1996 | dense | | 693 | ACF-WIDER (Chinese Academy of Sciences, 2014) | 6.1 × 10³ | 6.1 × 10³ | 2014 | dense | | 694 | RankNet (Microsoft Research, 2005) | 5.7 × 10³ | 5.7 × 10³ | 2005 | dense | | 695 | ANN Eye Tracker (, 1993) | 5.6 × 10³ | 5.6 × 10³ | 1993 | dense | | 696 | JPMAX (, 1994) | \~4.4 × 10³ | | 1994 | estimate | | 697 | Kohonen network (Helsinki University of Technology, 1981) | \~4.1 × 10³ | | 1981 | estimate | | 698 | Speaker-independent vowel classification (University of Washington, 1989) | 3 × 10³ | 3 × 10³ | 1989 | dense | | 699 | Pattern recognition and reading by machine (Sandia Corporation, 1959) | \~2.6 × 10³ | | 1959 | estimate | | 700 | Handwritten digit recognition network (AT&T, 1989) | 2.6 × 10³ | 2.6 × 10³ | 1989 | dense | | 701 | Boosting (Bell Laboratories, 1992) | \~2.6 × 10³ | | 1992 | estimate | | 702 | Decision tree adaline (Tokyo Medical and Dental University, 1969) | 2.5 × 10³ | 2.5 × 10³ | 1969 | dense | | 703 | LTE speaker verification system (IBM, 1966) | \~2.1 × 10³ | | 1966 | estimate | | 704 | Ceramic-MLP (Sapienza Università di Roma, 1994) | \~1.9 × 10³ | | 1994 | estimate | | 705 | SexNet classification (, 1990) | \~1.6 × 10³ | | 1990 | estimate | | 706 | RAAM (, 1990) | 1.5 × 10³ | 1.5 × 10³ | 1990 | dense | | 707 | Super-vector coding (University of Illinois Urbana-Champaign (UIUC), 2010) | \~1 × 10³ | | 2010 | estimate | | 708 | Perceptron Mark I (Cornell Aeronautical Laboratory, 1957) | \~1 × 10³ | | 1957 | estimate | | 709 | Perceptron (Cornell Aeronautical Laboratory, 1960) | \~1 × 10³ | | 1960 | estimate | | 710 | Translation-invariant MLP (Carnegie Mellon University (CMU), 1987) | 8.2 × 10² | 8.2 × 10² | 1987 | mixture of experts (active count unpublished) | | 711 | Truck backer-upper (Stanford University, 1989) | 8.1 × 10² | 8.1 × 10² | 1989 | dense | | 712 | Siamese-TDNN (Bell Laboratories, 1993) | \~7.4 × 10² | | 1993 | estimate | | 713 | MLP with back-propagation (University of California San Diego, 1986) | 7.2 × 10² | 7.2 × 10² | 1986 | dense | | 714 | Cancer drug mechanism prediction (National Cancer Institute, 1992) | \~5.9 × 10² | | 1992 | estimate | | 715 | Distributed representation NN (Carnegie Mellon University (CMU), 1986) | 4.3 × 10² | 4.3 × 10² | 1986 | dense | | 716 | Piecewise linear model (University of Kansas, 1973) | 3.6 × 10² | 3.6 × 10² | 1973 | dense | | 717 | ASE+ACE (University of Massachusetts Amherst, 1983) | \~3.2 × 10² | | 1983 | estimate | | 718 | LSTM with forget gates (IDSIA, 1999) | \~2.8 × 10² | | 1999 | estimate | | 719 | Self Organizing System (Massachusetts Institute of Technology (MIT), 1955) | \~2.3 × 10² | | 1955 | estimate | | 720 | Theseus (Bell Laboratories, 1950) | 4 × 10¹ | 4 × 10¹ | 1950 | dense | | 721 | SNARC (Harvard University, 1952) | \~4 × 10¹ | | 1952 | estimate | | 722 | Bankruptcy-NN (, 1990) | 3.6 × 10¹ | 3.6 × 10¹ | 1990 | dense | | 723 | GNN (University of Siena, 2008) | 3 × 10¹ | 3 × 10¹ | 2008 | dense | | 724 | ADALINE (Stanford University, 1960) | 1.7 × 10¹ | 1.7 × 10¹ | 1960 | dense | | 725 | Samuel Neural Checkers (IBM, 1959) | \~1.6 × 10¹ | | 1959 | estimate | ## Collectives These charts rank animal societies, religions, states, parties and corporations two ways. One is by members. The other is by neuron-equivalents, which weight each member by the share of their waking time the group’s instructions direct. Religious codes, laws, party rules and job descriptions all count the same way. They're all sets of instructions, so we measure them by the same rule. [How directed time is estimated](https://simulation-hypothesis.com/entities/control). ## Animal societies: neurons in coordinated group activity Neurons in the group In group activity Members of a typical group × neurons in each member’s cortex or pallium (whole brain for insects), on a logarithmic scale. Hatched bars use an estimated group size or a related species’ neuron count. 1. Chimpanzee community▲1 7.4 × 10¹¹ 2. Leafcutter ant colony▼1 6.3 × 10¹¹ 3. Bonobo community▲2 4.4 × 10¹¹ 4. Hadza camp (humans, for scale) 4.1 × 10¹¹ 5. Bigg’s orca group▼2 2.6 × 10¹¹ 6. Resident orca matriline 2.4 × 10¹¹ 7. Baboon troop 8.6 × 10¹⁰ 8. Bottlenose dolphin alliance▲1 7.6 × 10¹⁰ 9. Sperm whale social unit▲1 6.4 × 10¹⁰ 10. Honeybee colony▼2 5.8 × 10¹⁰ 11. Elephant family 5.6 × 10¹⁰ 12. Pinyon jay flock 5.3 × 10¹⁰ 13. Spotted hyena clan 2.2 × 10¹⁰ 14. African wild dog pack 7.2 × 10⁹ 15. Wolf pack 4.2 × 10⁹ 16. Meerkat group 1.7 × 10⁹ 17. Naked mole-rat colony 4.1 × 10⁸ 10^8 10^9 10^10 10^11 10^12 neurons Show the numbers | # | Group | Neurons in the group | In group activity | Members | Share directed | Call types | | -- | ------------------------------ | -------------------- | ----------------- | ----------- | -------------- | ---------- | | 1 | Chimpanzee community | 7.4 × 10¹¹ | 2.5 × 10¹¹ | 100 | 34% | — | | 2 | Leafcutter ant colony | 6.3 × 10¹¹ | 5 × 10¹¹ | 2.5 million | 80% | — | | 3 | Bonobo community | 4.4 × 10¹¹ | 1.9 × 10¹¹ | 59 | 43% | — | | 4 | Hadza camp (humans, for scale) | 4.1 × 10¹¹ | 2.1 × 10¹¹ | 25 | 53% | — | | 5 | Bigg’s orca group | 2.6 × 10¹¹ | 2.4 × 10¹¹ | 6 | 91% | — | | 6 | Resident orca matriline | 2.4 × 10¹¹ | 1.7 × 10¹¹ | 6 | 70% | 11 | | 7 | Baboon troop | 8.6 × 10¹⁰ | 5.2 × 10¹⁰ | 30 | 60% | 9 | | 8 | Bottlenose dolphin alliance | 7.6 × 10¹⁰ | 4.2 × 10¹⁰ | 6 | 55% | — | | 9 | Sperm whale social unit | 6.4 × 10¹⁰ | 4.1 × 10¹⁰ | 8 | 65% | — | | 10 | Honeybee colony | 5.8 × 10¹⁰ | 4.9 × 10¹⁰ | 60,000 | 85% | — | | 11 | Elephant family | 5.6 × 10¹⁰ | 3.2 × 10¹⁰ | 10 | 57% | — | | 12 | Pinyon jay flock | 5.3 × 10¹⁰ | 3.2 × 10¹⁰ | 100 | 60% | — | | 13 | Spotted hyena clan | 2.2 × 10¹⁰ | 6.7 × 10⁹ | 45 | 30% | 11 | | 14 | African wild dog pack | 7.2 × 10⁹ | 4.8 × 10⁹ | 14 | 67% | 11 | | 15 | Wolf pack | 4.2 × 10⁹ | 2.6 × 10⁹ | 8 | 63% | — | | 16 | Meerkat group | 1.7 × 10⁹ | 1 × 10⁹ | 15 | 57% | 12 | | 17 | Naked mole-rat colony | 4.1 × 10⁸ | 2.9 × 10⁸ | 78 | 70% | — | [How the groups are scored](https://simulation-hypothesis.com/entities/control). ## Religions: members, and neurons following their instructions Members Following its instructions Whole religions Sects and movements Adherents, on a logarithmic scale. Sects and movements are shown apart from the religions that contain them. Solid bars are counts; hatched bars are estimates. 1. Christianity▲1 2.27 billion 2. Islam▼1 2.02 billion 3. Hinduism 1.18 billion 4. Folk religions 400 million 5. Buddhism 324 million 6. African traditional religions▲1 98.7 million 7. Shinto▲1 87.9 million 8. Sikhism▲1 25 million 9. Judaism▼3 14.8 million 10. Taoism▲1 8 million 11. Baháʼí Faith▲1 8 million 12. Falun Gong▲1 7 million 13. Jainism▼3 4.45 million 14. Spiritism 3.8 million 15. Cao Đài 2.5 million 16. Tenrikyo 1.12 million 17. Brahma Kumaris 1 million 10^5 10^6 10^7 10^8 10^9 10^10 members Show the numbers | # | Religion | Members | Following its instructions | Share directed | Hours a week | | -- | ----------------------------- | ------------ | -------------------------- | -------------- | ------------ | | 1 | Christianity | 2.27 billion | 1.7 × 10¹⁸ | 0.88% | 0.9 h | | 2 | Islam | 2.02 billion | 8.9 × 10¹⁸ | 5.1% | 5.4 h | | 3 | Hinduism | 1.18 billion | 1.6 × 10¹⁸ | 1.6% | 1.6 h | | 4 | Folk religions | 400 million | 2 × 10¹⁷ | 0.59% (proxy) | 0.6 h | | 5 | Buddhism | 324 million | 7.5 × 10¹⁶ | 0.27% | 0.3 h | | 6 | African traditional religions | 98.7 million | 5 × 10¹⁶ | 0.59% (proxy) | 0.6 h | | 7 | Shinto | 87.9 million | 4.5 × 10¹⁶ | 0.59% (proxy) | 0.6 h | | 8 | Sikhism | 25 million | 3.7 × 10¹⁶ | 1.7% | 1.8 h | | 9 | Judaism | 14.8 million | 5.3 × 10¹⁶ | 4.2% | 4.4 h | | 10 | Taoism | 8 million | 4.1 × 10¹⁵ | 0.59% (proxy) | 0.6 h | | 11 | Baháʼí Faith | 8 million | 4.1 × 10¹⁵ | 0.59% (proxy) | 0.6 h | | 12 | Falun Gong | 7 million | 3.6 × 10¹⁵ | 0.59% (proxy) | 0.6 h | | 13 | Jainism | 4.45 million | 1.8 × 10¹⁶ | 4.6% | 4.8 h | | 14 | Spiritism | 3.8 million | 1.9 × 10¹⁵ | 0.59% (proxy) | 0.6 h | | 15 | Cao Đài | 2.5 million | 1.3 × 10¹⁵ | 0.59% (proxy) | 0.6 h | | 16 | Tenrikyo | 1.12 million | 5.7 × 10¹⁴ | 0.59% (proxy) | 0.6 h | | 17 | Brahma Kumaris | 1 million | 5.1 × 10¹⁴ | 0.59% (proxy) | 0.6 h | [Every practice, share and source](https://simulation-hypothesis.com/entities/control). ## States: residents, and neurons following their instructions Residents Following its instructions All East Asia, Pacific Europe, Central Asia Latin America Middle East, N. Africa North America South Asia Sub-Saharan Africa Population, on a logarithmic scale. Solid bars are counts; hatched bars are estimates. 1. India▲1 1.46 billion 2. China▼1 1.41 billion 3. United States 342 million 4. Indonesia 286 million 5. Pakistan▲5 255 million 6. Nigeria▲1 238 million 7. Brazil▼2 213 million 8. Bangladesh▲5 176 million 9. Russia▼3 144 million 10. Ethiopia▲7 135 million 11. Mexico▼2 132 million 12. Japan▲11 123 million 13. Egypt▼2 118 million 14. Philippines 117 million 15. DR Congo▼3 113 million Show all 196 10^4 10^6 10^8 10^10 residents Show the numbers | # | State | Residents | Following its instructions | Share directed | Hours a week | | --- | -------------------------------- | ------------ | -------------------------- | -------------- | ------------ | | 1 | India | 1.46 billion | 2.5 × 10¹⁸ | 2% | 2.1 h | | 2 | China | 1.41 billion | 5 × 10¹⁸ | 4.1% | 4.3 h | | 3 | United States | 342 million | 1.6 × 10¹⁸ | 5.6% | 5.8 h | | 4 | Indonesia | 286 million | 1 × 10¹⁸ | 4.2% | 4.4 h | | 5 | Pakistan | 255 million | 5.7 × 10¹⁷ | 2.6% | 2.7 h | | 6 | Nigeria | 238 million | 6.6 × 10¹⁷ | 3.2% | 3.4 h | | 7 | Brazil | 213 million | 8.7 × 10¹⁷ | 4.8% | 5 h | | 8 | Bangladesh | 176 million | 4.6 × 10¹⁷ | 3% | 3.1 h | | 9 | Russia | 144 million | 7.5 × 10¹⁷ | 6.1% | 6.4 h | | 10 | Ethiopia | 135 million | 3.9 × 10¹⁷ | 3.4% | 3.5 h | | 11 | Mexico | 132 million | 5.9 × 10¹⁷ | 5.2% | 5.5 h | | 12 | Japan | 123 million | 2.8 × 10¹⁷ | 2.7% | 2.8 h | | 13 | Egypt | 118 million | 5.5 × 10¹⁷ | 5.4% | 5.7 h | | 14 | Philippines | 117 million | 4.4 × 10¹⁷ | 4.4% | 4.6 h | | 15 | DR Congo | 113 million | 4.7 × 10¹⁷ | 4.8% | 5.1 h | | 16 | Vietnam | 102 million | 4.2 × 10¹⁷ | 4.8% | 5 h | | 17 | Iran | 92.4 million | 3.5 × 10¹⁷ | 4.3% | 4.6 h | | 18 | Turkey | 85.9 million | 4.3 × 10¹⁷ | 5.8% | 6 h | | 19 | Germany | 83.5 million | 2.9 × 10¹⁷ | 4% | 4.2 h | | 20 | Thailand | 71.6 million | 2.4 × 10¹⁷ | 3.9% | 4.1 h | | 21 | Tanzania | 70.5 million | 2.2 × 10¹⁷ | 3.6% | 3.8 h | | 22 | United Kingdom | 69.5 million | 2.8 × 10¹⁷ | 4.6% | 4.8 h | | 23 | France | 68.7 million | 3.1 × 10¹⁷ | 5.3% | 5.5 h | | 24 | South Africa | 64.7 million | 2.9 × 10¹⁷ | 5.1% | 5.4 h | | 25 | Italy | 58.9 million | 2.1 × 10¹⁷ | 4.2% | 4.4 h | | 26 | Kenya | 57.5 million | 2.3 × 10¹⁷ | 4.7% | 5 h | | 27 | Myanmar | 54.9 million | 1.9 × 10¹⁷ | 4% | 4.2 h | | 28 | Colombia | 53.4 million | 1.7 × 10¹⁷ | 3.7% | 3.9 h | | 29 | South Korea | 51.7 million | 1.4 × 10¹⁷ | 3% | 3.2 h | | 30 | Sudan | 51.7 million | 1.5 × 10¹⁷ | 3.4% | 3.6 h | | 31 | Uganda | 51.4 million | 1.5 × 10¹⁷ | 3.4% | 3.6 h | | 32 | Spain | 49.4 million | 1.8 × 10¹⁷ | 4.2% | 4.3 h | | 33 | Algeria | 47.4 million | 3.2 × 10¹⁷ | 7.9% | 8.3 h | | 34 | Iraq | 47 million | 1.1 × 10¹⁷ | 2.8% | 2.9 h | | 35 | Argentina | 45.9 million | 1.7 × 10¹⁷ | 4.4% | 4.6 h | | 36 | Afghanistan | 43.8 million | 1.8 × 10¹⁷ | 4.8% | 5 h | | 37 | Yemen | 41.8 million | 1.3 × 10¹⁷ | 3.7% | 3.9 h | | 38 | Canada | 41.7 million | 2 × 10¹⁷ | 5.5% | 5.8 h | | 39 | Angola | 39 million | 1.4 × 10¹⁷ | 4.2% | 4.4 h | | 40 | Ukraine | 39 million | 2.1 × 10¹⁷ | 6.2% | 6.5 h | | 41 | Morocco | 38.4 million | 1.5 × 10¹⁷ | 4.5% | 4.7 h | | 42 | Uzbekistan | 37.1 million | 2 × 10¹⁷ | 6.3% | 6.6 h | | 43 | Saudi Arabia | 37 million | 1.4 × 10¹⁷ | 4.3% | 4.6 h | | 44 | Poland | 36.4 million | 2.1 × 10¹⁷ | 6.7% | 7 h | | 45 | Malaysia | 36 million | 1.8 × 10¹⁷ | 5.8% | 6.1 h | | 46 | Mozambique | 35.6 million | 1.5 × 10¹⁷ | 4.8% | 5.1 h | | 47 | Ghana | 35.1 million | 1.2 × 10¹⁷ | 3.9% | 4.1 h | | 48 | Peru | 34.6 million | 1.4 × 10¹⁷ | 4.8% | 5 h | | 49 | Madagascar | 32.7 million | 9.4 × 10¹⁶ | 3.3% | 3.5 h | | 50 | Côte d’Ivoire | 32.7 million | 9.5 × 10¹⁶ | 3.4% | 3.5 h | | 51 | Cameroon | 29.9 million | 9.8 × 10¹⁶ | 3.8% | 4 h | | 52 | Nepal | 29.6 million | 1.1 × 10¹⁷ | 4.2% | 4.4 h | | 53 | Venezuela | 28.5 million | 1.5 × 10¹⁷ | 6.1% | 6.4 h | | 54 | Niger | 27.9 million | 6.4 × 10¹⁶ | 2.6% | 2.8 h | | 55 | Australia | 27.6 million | 1.1 × 10¹⁷ | 4.7% | 5 h | | 56 | North Korea | 26.6 million | 6.6 × 10¹⁷ | 29% | 30.1 h | | 57 | Syria | 25.6 million | 9.8 × 10¹⁶ | 4.4% | 4.6 h | | 58 | Mali | 25.2 million | 4.9 × 10¹⁶ | 2.3% | 2.4 h | | 59 | Burkina Faso | 24.1 million | 5.7 × 10¹⁶ | 2.8% | 2.9 h | | 60 | Taiwan | 23.3 million | 1 × 10¹⁷ | 5% | 5.3 h | | 61 | Malawi | 22.2 million | 1 × 10¹⁷ | 5.4% | 5.6 h | | 62 | Zambia | 21.9 million | 10 × 10¹⁶ | 5.3% | 5.5 h | | 63 | Sri Lanka | 21.8 million | 1.1 × 10¹⁷ | 6% | 6.3 h | | 64 | Chad | 21 million | 4.9 × 10¹⁶ | 2.7% | 2.9 h | | 65 | Kazakhstan | 20.8 million | 1.4 × 10¹⁷ | 7.6% | 7.9 h | | 66 | Chile | 19.9 million | 6.2 × 10¹⁶ | 3.6% | 3.8 h | | 67 | Somalia | 19.7 million | 1.5 × 10¹⁶ | 0.88% | 0.9 h | | 68 | Romania | 19 million | 7.5 × 10¹⁶ | 4.6% | 4.8 h | | 69 | Senegal | 18.9 million | 5.9 × 10¹⁶ | 3.6% | 3.8 h | | 70 | Guatemala | 18.7 million | 6 × 10¹⁶ | 3.7% | 3.9 h | | 71 | Ecuador | 18.3 million | 6.6 × 10¹⁶ | 4.2% | 4.4 h | | 72 | Netherlands | 18.1 million | 7.6 × 10¹⁶ | 4.9% | 5.1 h | | 73 | Cambodia | 17.8 million | 7.4 × 10¹⁶ | 4.8% | 5.1 h | | 74 | Zimbabwe | 17 million | 6.1 × 10¹⁶ | 4.2% | 4.4 h | | 75 | Guinea | 15.1 million | 3.1 × 10¹⁶ | 2.4% | 2.5 h | | 76 | Benin | 14.8 million | 4.4 × 10¹⁶ | 3.4% | 3.6 h | | 77 | Rwanda | 14.6 million | 7 × 10¹⁶ | 5.6% | 5.8 h | | 78 | Burundi | 14.4 million | 5.2 × 10¹⁶ | 4.2% | 4.4 h | | 79 | Bolivia | 12.6 million | 5.8 × 10¹⁶ | 5.4% | 5.6 h | | 80 | Tunisia | 12.3 million | 6.4 × 10¹⁶ | 6% | 6.3 h | | 81 | South Sudan | 12.2 million | 4.4 × 10¹⁶ | 4.2% | 4.4 h | | 82 | Belgium | 11.9 million | 4.5 × 10¹⁶ | 4.3% | 4.5 h | | 83 | Haiti | 11.9 million | 1.9 × 10¹⁶ | 1.8% | 1.9 h | | 84 | Jordan | 11.5 million | 4.9 × 10¹⁶ | 5% | 5.2 h | | 85 | Dominican Republic | 11.5 million | 4.9 × 10¹⁶ | 4.9% | 5.2 h | | 86 | United Arab Emirates | 11.5 million | 3.1 × 10¹⁶ | 3.1% | 3.2 h | | 87 | Honduras | 11 million | 3.2 × 10¹⁶ | 3.3% | 3.5 h | | 88 | Cuba | 10.9 million | 1.6 × 10¹⁷ | 17% | 17.4 h | | 89 | Czechia | 10.9 million | 5.8 × 10¹⁶ | 6.2% | 6.5 h | | 90 | Portugal | 10.8 million | 5.1 × 10¹⁶ | 5.5% | 5.7 h | | 91 | Tajikistan | 10.8 million | 5.3 × 10¹⁶ | 5.7% | 6 h | | 92 | Papua New Guinea | 10.8 million | 4.3 × 10¹⁶ | 4.6% | 4.8 h | | 93 | Sweden | 10.6 million | 6.8 × 10¹⁶ | 7.5% | 7.8 h | | 94 | Greece | 10.4 million | 4.9 × 10¹⁶ | 5.4% | 5.7 h | | 95 | Azerbaijan | 10.2 million | 6.2 × 10¹⁶ | 7% | 7.4 h | | 96 | Israel | 10.1 million | 6.4 × 10¹⁶ | 7.3% | 7.7 h | | 97 | Hungary | 9.51 million | 5.7 × 10¹⁶ | 6.9% | 7.3 h | | 98 | Austria | 9.21 million | 2.5 × 10¹⁶ | 3.2% | 3.4 h | | 99 | Switzerland | 9.09 million | 3.9 × 10¹⁶ | 5% | 5.3 h | | 100 | Belarus | 9.09 million | 7.9 × 10¹⁶ | 10% | 10.6 h | | 101 | Sierra Leone | 8.82 million | 4.9 × 10¹⁶ | 6.5% | 6.8 h | | 102 | Togo | 8.59 million | 3 × 10¹⁶ | 4% | 4.2 h | | 103 | Laos | 7.87 million | 3.2 × 10¹⁶ | 4.8% | 5 h | | 104 | Turkmenistan | 7.62 million | 4.2 × 10¹⁶ | 6.5% | 6.8 h | | 105 | Libya | 7.46 million | 3.9 × 10¹⁶ | 6.1% | 6.4 h | | 106 | Kyrgyzstan | 7.34 million | 4.1 × 10¹⁶ | 6.5% | 6.8 h | | 107 | Paraguay | 7.01 million | 2.9 × 10¹⁶ | 4.8% | 5.1 h | | 108 | Nicaragua | 7.01 million | 2.8 × 10¹⁶ | 4.6% | 4.8 h | | 109 | Serbia | 6.55 million | 4.1 × 10¹⁶ | 7.2% | 7.5 h | | 110 | Republic of the Congo | 6.48 million | 1.7 × 10¹⁶ | 3% | 3.1 h | | 111 | Bulgaria | 6.43 million | 3.3 × 10¹⁶ | 5.9% | 6.2 h | | 112 | El Salvador | 6.37 million | 3.2 × 10¹⁶ | 5.8% | 6.1 h | | 113 | Singapore | 6.11 million | 1.9 × 10¹⁶ | 3.6% | 3.7 h | | 114 | Denmark | 6.01 million | 3.9 × 10¹⁶ | 7.6% | 8 h | | 115 | Lebanon | 5.85 million | 1.5 × 10¹⁶ | 3% | 3.1 h | | 116 | Liberia | 5.73 million | 1.5 × 10¹⁶ | 3.1% | 3.3 h | | 117 | Finland | 5.65 million | 3.2 × 10¹⁶ | 6.6% | 6.9 h | | 118 | Norway | 5.61 million | 3.7 × 10¹⁶ | 7.6% | 8 h | | 119 | Central African Republic | 5.51 million | 1.7 × 10¹⁶ | 3.5% | 3.7 h | | 120 | Oman | 5.49 million | 2.9 × 10¹⁶ | 6.2% | 6.5 h | | 121 | Ireland | 5.48 million | 3.4 × 10¹⁶ | 7.2% | 7.6 h | | 122 | Slovakia | 5.41 million | 3.1 × 10¹⁶ | 6.7% | 7 h | | 123 | Palestine | 5.41 million | 2 × 10¹⁶ | 4.3% | 4.5 h | | 124 | New Zealand | 5.32 million | 2.7 × 10¹⁶ | 5.9% | 6.2 h | | 125 | Mauritania | 5.32 million | 2 × 10¹⁶ | 4.3% | 4.5 h | | 126 | Costa Rica | 5.15 million | 2.6 × 10¹⁶ | 5.9% | 6.2 h | | 127 | Kuwait | 4.87 million | 2.7 × 10¹⁶ | 6.5% | 6.8 h | | 128 | Panama | 4.57 million | 2.3 × 10¹⁶ | 5.9% | 6.2 h | | 129 | Georgia | 3.94 million | 2.1 × 10¹⁶ | 6.1% | 6.4 h | | 130 | Croatia | 3.88 million | 2.2 × 10¹⁶ | 6.7% | 7 h | | 131 | Eritrea | 3.61 million | 1.5 × 10¹⁶ | 5% | 5.2 h | | 132 | Mongolia | 3.57 million | 2.2 × 10¹⁶ | 7% | 7.3 h | | 133 | Uruguay | 3.38 million | 1.7 × 10¹⁶ | 5.9% | 6.2 h | | 134 | Bosnia and Herzegovina | 3.14 million | 1.6 × 10¹⁶ | 5.9% | 6.2 h | | 135 | Namibia | 3.09 million | 1.8 × 10¹⁶ | 6.9% | 7.2 h | | 136 | Armenia | 3.09 million | 1.7 × 10¹⁶ | 6.5% | 6.8 h | | 137 | Qatar | 2.97 million | 9.6 × 10¹⁵ | 3.8% | 3.9 h | | 138 | Lithuania | 2.89 million | 1.8 × 10¹⁶ | 7.4% | 7.7 h | | 139 | Jamaica | 2.84 million | 1.2 × 10¹⁶ | 5% | 5.2 h | | 140 | Gambia | 2.82 million | 1.1 × 10¹⁶ | 4.5% | 4.7 h | | 141 | Gabon | 2.59 million | 9.9 × 10¹⁵ | 4.4% | 4.6 h | | 142 | Botswana | 2.56 million | 1.3 × 10¹⁶ | 5.7% | 6 h | | 143 | Lesotho | 2.36 million | 9.5 × 10¹⁵ | 4.7% | 4.9 h | | 144 | Moldova | 2.36 million | 1.5 × 10¹⁶ | 7.3% | 7.6 h | | 145 | Albania | 2.35 million | 1.2 × 10¹⁶ | 5.7% | 6 h | | 146 | Guinea-Bissau | 2.25 million | 5.7 × 10¹⁵ | 2.9% | 3.1 h | | 147 | Slovenia | 2.13 million | 1.2 × 10¹⁶ | 6.6% | 6.9 h | | 148 | Equatorial Guinea | 1.94 million | 4 × 10¹⁵ | 2.4% | 2.5 h | | 149 | Latvia | 1.85 million | 1.2 × 10¹⁶ | 7.6% | 7.9 h | | 150 | North Macedonia | 1.82 million | 9.2 × 10¹⁵ | 5.9% | 6.2 h | | 151 | Bahrain | 1.6 million | 5.9 × 10¹⁵ | 4.3% | 4.5 h | | 152 | Kosovo | 1.58 million | 8 × 10¹⁵ | 5.9% | 6.2 h | | 153 | Timor-Leste | 1.42 million | 8 × 10¹⁵ | 6.6% | 6.9 h | | 154 | Cyprus | 1.37 million | 5.5 × 10¹⁵ | 4.7% | 4.9 h | | 155 | Trinidad and Tobago | 1.37 million | 9.3 × 10¹⁵ | 7.9% | 8.2 h | | 156 | Estonia | 1.37 million | 8.3 × 10¹⁵ | 7% | 7.4 h | | 157 | Eswatini | 1.26 million | 7.5 × 10¹⁵ | 6.9% | 7.2 h | | 158 | Mauritius | 1.24 million | 5.1 × 10¹⁵ | 4.8% | 5 h | | 159 | Djibouti | 1.18 million | 4.6 × 10¹⁵ | 4.5% | 4.8 h | | 160 | Fiji | 933,154 | 5.6 × 10¹⁵ | 7% | 7.4 h | | 161 | Comoros | 882,847 | 3.5 × 10¹⁵ | 4.6% | 4.8 h | | 162 | Solomon Islands | 838,645 | 3.2 × 10¹⁵ | 4.4% | 4.6 h | | 163 | Guyana | 835,986 | 4.6 × 10¹⁵ | 6.4% | 6.7 h | | 164 | Bhutan | 796,682 | 5.8 × 10¹⁵ | 8.5% | 8.9 h | | 165 | Luxembourg | 686,970 | 2.4 × 10¹⁵ | 4% | 4.2 h | | 166 | Suriname | 639,850 | 3.7 × 10¹⁵ | 6.7% | 7.1 h | | 167 | Montenegro | 623,129 | 4.3 × 10¹⁵ | 8.1% | 8.5 h | | 168 | Malta | 579,704 | 3.1 × 10¹⁵ | 6.2% | 6.5 h | | 169 | Maldives | 529,676 | 4.7 × 10¹⁵ | 10% | 10.8 h | | 170 | Cape Verde | 527,326 | 3.3 × 10¹⁵ | 7.2% | 7.6 h | | 171 | Brunei | 466,330 | 3.6 × 10¹⁵ | 8.9% | 9.3 h | | 172 | Belize | 422,924 | 1.1 × 10¹⁵ | 2.9% | 3.1 h | | 173 | Bahamas | 403,033 | 2.2 × 10¹⁵ | 6.4% | 6.7 h | | 174 | Iceland | 392,404 | 2.5 × 10¹⁵ | 7.4% | 7.7 h | | 175 | Vanuatu | 335,169 | 2 × 10¹⁵ | 6.9% | 7.3 h | | 176 | Barbados | 282,623 | 1.5 × 10¹⁵ | 6% | 6.3 h | | 177 | São Tomé and Príncipe | 240,254 | 1.1 × 10¹⁵ | 5.4% | 5.7 h | | 178 | Samoa | 219,306 | 1.1 × 10¹⁵ | 5.7% | 5.9 h | | 179 | Saint Lucia | 180,149 | 9.3 × 10¹⁴ | 6% | 6.3 h | | 180 | Kiribati | 136,488 | 7.6 × 10¹⁴ | 6.4% | 6.7 h | | 181 | Seychelles | 122,730 | 8.9 × 10¹⁴ | 8.4% | 8.8 h | | 182 | Grenada | 117,303 | 7.3 × 10¹⁴ | 7.2% | 7.5 h | | 183 | Micronesia | 113,683 | 5.7 × 10¹⁴ | 5.8% | 6.1 h | | 184 | Tonga | 103,742 | 5.9 × 10¹⁴ | 6.6% | 6.9 h | | 185 | Saint Vincent and the Grenadines | 99,924 | 4.9 × 10¹⁴ | 5.7% | 5.9 h | | 186 | Antigua and Barbuda | 94,209 | 6.5 × 10¹⁴ | 8% | 8.3 h | | 187 | Andorra | 82,904 | 3.9 × 10¹⁴ | 5.5% | 5.8 h | | 188 | Dominica | 65,871 | 2.3 × 10¹⁴ | 4.1% | 4.3 h | | 189 | Saint Kitts and Nevis | 46,922 | 2.1 × 10¹⁴ | 5.2% | 5.5 h | | 190 | Liechtenstein | 41,024 | 2 × 10¹⁴ | 5.7% | 5.9 h | | 191 | Monaco | 38,341 | 2.3 × 10¹⁴ | 6.9% | 7.3 h | | 192 | Marshall Islands | 36,282 | 3.3 × 10¹⁴ | 10% | 10.9 h | | 193 | San Marino | 34,109 | 1.3 × 10¹⁴ | 4.6% | 4.8 h | | 194 | Palau | 17,663 | 1.6 × 10¹⁴ | 10% | 11 h | | 195 | Nauru | 12,025 | 9.3 × 10¹³ | 9% | 9.4 h | | 196 | Tuvalu | 9,492 | 8 × 10¹³ | 9.8% | 10.2 h | [What’s counted, and what isn’t](https://simulation-hypothesis.com/entities/control). ## Political parties: members, and neurons following their instructions Members Following its instructions Members as the party reports them, on a logarithmic scale. Solid bars are counts; hatched bars are estimates or claims. 1. Bharatiya Janata Party (India)▲1 140 million 2. Chinese Communist Party▼1 101 million 3. Democratic Party (US)▲13 44.9 million 4. Republican Party (US)▲16 38 million 5. Indian National Congress▼2 26 million 6. AIADMK (India)▼2 20 million 7. DMK (India)▼2 20 million 8. Prosperity Party (Ethiopia)▼1 15.7 million 9. TVK (India)▼1 15 million 10. Chama Cha Mapinduzi (Tanzania)▼1 13 million 11. All Progressives Congress (Nigeria)▼1 12.9 million 12. AK Party (Turkey)▼1 11.7 million 13. Morena (Mexico)▼1 11.6 million 14. Telugu Desam Party (India)▼1 10.1 million 15. Cambodian People’s Party▼1 7 million Show all 59 10^4 10^5 10^6 10^7 10^8 10^9 members Show the numbers | # | Party | Members | Following its instructions | Share directed | Hours a week | | -- | ------------------------------------------ | ------------ | -------------------------- | -------------- | ------------ | | 1 | Bharatiya Janata Party (India) | 140 million | 6.3 × 10¹⁶ | 0.52% (proxy) | 0.6 h | | 2 | Chinese Communist Party | 101 million | 8.2 × 10¹⁶ | 0.94% | 1 h | | 3 | Democratic Party (US) | 44.9 million | 2.3 × 10¹⁵ | 0.06% | 0.1 h | | 4 | Republican Party (US) | 38 million | 2 × 10¹⁵ | 0.06% | 0.1 h | | 5 | Indian National Congress | 26 million | 1.2 × 10¹⁶ | 0.52% (proxy) | 0.6 h | | 6 | AIADMK (India) | 20 million | 9 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 7 | DMK (India) | 20 million | 9 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 8 | Prosperity Party (Ethiopia) | 15.7 million | 7 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 9 | TVK (India) | 15 million | 6.7 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 10 | Chama Cha Mapinduzi (Tanzania) | 13 million | 5.8 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 11 | All Progressives Congress (Nigeria) | 12.9 million | 5.8 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 12 | AK Party (Turkey) | 11.7 million | 5.2 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 13 | Morena (Mexico) | 11.6 million | 5.2 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 14 | Telugu Desam Party (India) | 10.1 million | 4.5 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 15 | Cambodian People’s Party | 7 million | 3.1 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 16 | Workers’ Party of Korea | 6.17 million | 7.6 × 10¹⁵ | 1.4% | 1.5 h | | 17 | FRELIMO (Mozambique) | 6 million | 2.7 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 18 | Communist Party of Vietnam | 5.6 million | 2.1 × 10¹⁵ | 0.44% | 0.5 h | | 19 | Democratic Party of Korea | 5 million | 2.2 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 20 | People Power Party (South Korea) | 4.44 million | 2 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 21 | PSUV (Venezuela) | 4.24 million | 1.9 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 22 | ZANU-PF (Zimbabwe) | 4.06 million | 1.8 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 23 | UMNO (Malaysia) | 3.25 million | 1.5 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 24 | Justicialist Party (Argentina) | 3.1 million | 1.4 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 25 | Colorado Party (Paraguay) | 3 million | 1.3 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 26 | United Russia | 2.6 million | 1.2 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 27 | UzLiDeP (Uzbekistan) | 2.31 million | 1 × 10¹⁵ | 0.52% (proxy) | 0.6 h | | 28 | PLD (Dominican Republic) | 2.18 million | 9.7 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 29 | MDB (Brazil) | 2.02 million | 9 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 30 | CHP (Turkey) | 1.88 million | 8.4 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 31 | Radical Civic Union (Argentina) | 1.79 million | 8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 32 | Workers’ Party, PT (Brazil) | 1.67 million | 7.5 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 33 | PRM (Dominican Republic) | 1.55 million | 6.9 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 34 | Authentic Radical Liberal Party (Paraguay) | 1.55 million | 6.9 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 35 | Partido Federal ng Pilipinas | 1.5 million | 6.7 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 36 | PAS (Malaysia) | 1.3 million | 5.8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 37 | PRD (Brazil) | 1.29 million | 5.8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 38 | Progressistas (Brazil) | 1.29 million | 5.8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 39 | PSDB (Brazil) | 1.26 million | 5.6 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 40 | PKR (Malaysia) | 1.15 million | 5.2 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 41 | Economic Freedom Fighters (South Africa) | 1.09 million | 4.9 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 42 | PDT (Brazil) | 1.07 million | 4.8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 43 | União Brasil | 1.07 million | 4.8 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 44 | CPI(M) (India) | 1.02 million | 4.6 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 45 | Liberal Democratic Party (Japan) | 1 million | 4.5 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 46 | FLN (Algeria) | 1 million | 4.5 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 47 | Liberal Party, PL (Brazil) | 952,881 | 4.3 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 48 | Communist Party of Cuba | 700,000 | 2.7 × 10¹⁴ | 0.44% | 0.5 h | | 49 | African National Congress | 688,454 | 3.1 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 50 | Conservative Party of Canada | 678,708 | 2.7 × 10¹⁴ | 0.46% | 0.5 h | | 51 | CDU (Germany) | 364,202 | 1.6 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 52 | SPD (Germany) | 357,116 | 1.6 × 10¹⁴ | 0.52% (proxy) | 0.6 h | | 53 | Labour Party (UK) | 333,235 | 1.6 × 10¹⁴ | 0.55% | 0.6 h | | 54 | Liberal Party of Canada | 163,836 | 6.5 × 10¹³ | 0.46% | 0.5 h | | 55 | Greens (Germany) | 155,296 | 7 × 10¹³ | 0.52% (proxy) | 0.6 h | | 56 | CSU (Germany) | 125,153 | 5.6 × 10¹³ | 0.52% (proxy) | 0.6 h | | 57 | Conservative Party (UK) | 123,000 | 3.7 × 10¹³ | 0.35% | 0.4 h | | 58 | New Democratic Party (Canada) | 100,542 | 4 × 10¹³ | 0.46% | 0.5 h | | 59 | AfD (Germany) | 51,335 | 2.3 × 10¹³ | 0.52% (proxy) | 0.6 h | [Meetings, surveys and sources](https://simulation-hypothesis.com/entities/control). ## Corporations: employees, and neurons following their instructions Employees Following its instructions Employees in the latest annual report or list, on a linear scale. 1. Walmart 2.1 million 2. Amazon▲1 1.58 million 3. State Grid▼1 1.36 million 4. China National Petroleum 1.03 million 5. JD.com 951,420 6. BYD▲1 869,622 7. Accenture▼1 779,000 8. Allied Universal▲3 770,000 9. China Post Group▼1 719,448 10. Russian Railways (RZD)▲2 677,363 11. Volkswagen Group▲6 662,942 12. Foxconn (Hon Hai)▼2 633,167 13. Schwarz Group▲16 604,000 14. Compass Group▲18 590,000 15. Tata Consultancy Services▼6 584,510 Show all 71 0 0.5M 1M 1.5M 2M 2.5M employees Show the numbers | # | Corporation | Employees | Following its instructions | Hours a week | Share directed | | -- | ------------------------------------ | ------------ | -------------------------- | ------------------------ | -------------- | | 1 | Walmart | 2.1 million | 6.3 × 10¹⁶ | 36.3 h (United States) | 35% | | 2 | Amazon | 1.58 million | 4.7 × 10¹⁶ | 36.3 h (United States) | 35% | | 3 | State Grid | 1.36 million | 5.5 × 10¹⁶ | 49 h (China) | 47% | | 4 | China National Petroleum | 1.03 million | 4.1 × 10¹⁶ | 49 h (China) | 47% | | 5 | JD.com | 951,420 | 3.8 × 10¹⁶ | 49 h (China) | 47% | | 6 | BYD | 869,622 | 3.5 × 10¹⁶ | 49 h (China) | 47% | | 7 | Accenture | 779,000 | 3.7 × 10¹⁶ | 57.58 h (India) | 55% | | 8 | Allied Universal | 770,000 | 2.3 × 10¹⁶ | 36.52 h (United States) | 35% | | 9 | China Post Group | 719,448 | 2.9 × 10¹⁶ | 49 h (China) | 47% | | 10 | Russian Railways (RZD) | 677,363 | 2.2 × 10¹⁶ | 39.96 h (Russia) | 38% | | 11 | Volkswagen Group | 662,942 | 2 × 10¹⁶ | 35.95 h (Germany) | 34% | | 12 | Foxconn (Hon Hai) | 633,167 | 2.5 × 10¹⁶ | 49 h (China) | 47% | | 13 | Schwarz Group | 604,000 | 1.6 × 10¹⁶ | 31.49 h (Germany) | 30% | | 14 | Compass Group | 590,000 | 1.5 × 10¹⁶ | 31.7 h (United States) | 30% | | 15 | Tata Consultancy Services | 584,510 | 2.8 × 10¹⁶ | 58.23 h (India) | 56% | | 16 | DHL Group | 583,998 | 1.7 × 10¹⁶ | 35.81 h (Germany) | 34% | | 17 | Sinopec | 513,434 | 2.1 × 10¹⁶ | 49 h (China) | 47% | | 18 | Gazprom | 501,000 | 1.7 × 10¹⁶ | 40.47 h (Russia) | 39% | | 19 | FedEx | 474,200 | 1.6 × 10¹⁶ | 39.98 h (United States) | 38% | | 20 | Home Depot | 470,000 | 1.4 × 10¹⁶ | 36.3 h (United States) | 35% | | 21 | China Mobile | 468,496 | 1.9 × 10¹⁶ | 49 h (China) | 47% | | 22 | UPS | 460,000 | 1.5 × 10¹⁶ | 39.98 h (United States) | 38% | | 23 | Agricultural Bank of China | 457,835 | 1.8 × 10¹⁶ | 49 h (China) | 47% | | 24 | Concentrix | 455,000 | 1.7 × 10¹⁶ | 46.71 h (Philippines) | 45% | | 25 | Teleperformance | 446,716 | 2.2 × 10¹⁶ | 58.89 h (India) | 56% | | 26 | Jardine Matheson | 443,000 | 1.5 × 10¹⁶ | 41.93 h (Indonesia) | 40% | | 27 | Sodexo | 426,464 | 1.1 × 10¹⁶ | 31.7 h (United States) | 30% | | 28 | Capgemini | 423,400 | 2 × 10¹⁶ | 58.23 h (India) | 56% | | 29 | Reliance Industries | 419,911 | 2.1 × 10¹⁶ | 59.57 h (India) | 57% | | 30 | Edeka | 417,500 | 1.1 × 10¹⁶ | 31.49 h (Germany) | 30% | | 31 | Luxshare Precision | 416,197 | 1.7 × 10¹⁶ | 49 h (China) | 47% | | 32 | Target | 415,000 | 1.2 × 10¹⁶ | 36.3 h (United States) | 35% | | 33 | Marriott International | 414,000 | 1.1 × 10¹⁶ | 31.7 h (United States) | 30% | | 34 | Bosch | 412,400 | 1.2 × 10¹⁶ | 35.95 h (Germany) | 34% | | 35 | ICBC | 409,758 | 1.6 × 10¹⁶ | 49 h (China) | 47% | | 36 | Kroger | 403,000 | 1.2 × 10¹⁶ | 36.3 h (United States) | 35% | | 37 | China Resources | 397,398 | 1.6 × 10¹⁶ | 49 h (China) | 47% | | 38 | Magnit | 394,000 | 1.3 × 10¹⁶ | 39.69 h (Russia) | 38% | | 39 | Toyota | 390,927 | 1.3 × 10¹⁶ | 39.6 h (Japan) | 38% | | 40 | UnitedHealth Group | 390,000 | 1.2 × 10¹⁶ | 36.87 h (United States) | 35% | | 41 | X5 Group | 389,000 | 1.3 × 10¹⁶ | 39.69 h (Russia) | 38% | | 42 | Berkshire Hathaway | 387,800 | 1.2 × 10¹⁶ | 37.75 h (United States) | 36% | | 43 | AVIC | 384,000 | 1.5 × 10¹⁶ | 49 h (China) | 47% | | 44 | Ahold Delhaize | 384,000 | 1.1 × 10¹⁶ | 36.3 h (United States) | 35% | | 45 | China Telecom | 381,436 | 1.5 × 10¹⁶ | 49 h (China) | 47% | | 46 | Starbucks | 381,000 | 9.9 × 10¹⁵ | 31.7 h (United States) | 30% | | 47 | Rewe Group | 380,000 | 9.8 × 10¹⁵ | 31.49 h (Germany) | 30% | | 48 | China Construction Bank | 378,344 | 1.5 × 10¹⁶ | 49 h (China) | 47% | | 49 | TJX Companies | 377,000 | 1.1 × 10¹⁶ | 36.3 h (United States) | 35% | | 50 | China State Construction Engineering | 370,272 | 1.5 × 10¹⁶ | 49 h (China) | 47% | | 51 | FEMSA | 368,776 | 1.4 × 10¹⁶ | 47.09 h (Mexico) | 45% | | 52 | Japan Post Holdings | 360,000 | 1.2 × 10¹⁶ | 41.6 h (Japan) | 40% | | 53 | Tesco | 354,744 | 9.3 × 10¹⁵ | 31.81 h (United Kingdom) | 30% | | 54 | Cognizant | 351,600 | 1.7 × 10¹⁶ | 58.23 h (India) | 56% | | 55 | NTT | 344,196 | 1.1 × 10¹⁶ | 40.2 h (Japan) | 38% | | 56 | Costco | 341,000 | 1 × 10¹⁶ | 36.3 h (United States) | 35% | | 57 | Infosys | 328,594 | 1.6 × 10¹⁶ | 58.23 h (India) | 56% | | 58 | ISS | 325,000 | 1.6 × 10¹⁶ | 58.89 h (India) | 56% | | 59 | Securitas | 322,000 | 9.7 × 10¹⁵ | 36.52 h (United States) | 35% | | 60 | HCA Healthcare | 320,000 | 9.7 × 10¹⁵ | 36.87 h (United States) | 35% | | 61 | Rosneft | 320,000 | 1.1 × 10¹⁶ | 40.47 h (Russia) | 39% | | 62 | JPMorgan Chase | 318,512 | 1.1 × 10¹⁶ | 40.14 h (United States) | 38% | | 63 | Siemens | 318,000 | 9.4 × 10¹⁵ | 35.95 h (Germany) | 34% | | 64 | Bank of China | 313,746 | 1.3 × 10¹⁶ | 49 h (China) | 47% | | 65 | Walgreens Boots Alliance | 312,000 | 9.3 × 10¹⁵ | 36.3 h (United States) | 35% | | 66 | PepsiCo | 306,000 | 1 × 10¹⁶ | 40.7 h (United States) | 39% | | 67 | China Railway Construction (CRCC) | 304,927 | 1.2 × 10¹⁶ | 49 h (China) | 47% | | 68 | CHN Energy (China Energy Investment) | 303,418 | 1.2 × 10¹⁶ | 49 h (China) | 47% | | 69 | China Railway Group (CREC) | 302,527 | 1.2 × 10¹⁶ | 49 h (China) | 47% | | 70 | CK Hutchison | 300,000 | 8.6 × 10¹⁵ | 35.08 h (United Kingdom) | 33% | | 71 | CVS Health | 300,000 | 8.9 × 10¹⁵ | 36.3 h (United States) | 35% | [Hours and sources](https://simulation-hypothesis.com/entities/control). ## Animals and other organisms From single cells with no neurons to the human brain. Neuron counts are measured where they exist. ### No neurons * [Escherichia coli](https://simulation-hypothesis.com/entities/ecoli)Escherichia coli 0 measured A single cell with no neurons. It steers toward food and has a chemical memory that lasts a few seconds. * [Slime mould](https://simulation-hypothesis.com/entities/physarum)Physarum polycephalum 0 measured A single giant cell with no neurons. It solves mazes and grows efficient transport networks. * [Sensitive plant](https://simulation-hypothesis.com/entities/mimosa)Mimosa pudica 0 measured A plant whose leaves fold when touched. One study reported that it learns to ignore harmless touches. A later re-analysis disputed that. * [Sponge](https://simulation-hypothesis.com/entities/sponge)Porifera (e.g. Amphimedon queenslandica) 0 measured An animal with no neurons or synapses that still contracts its whole body in a coordinated way. ### Nerve nets, no brain * [Hydra](https://simulation-hypothesis.com/entities/hydra)Hydra vulgaris about 3,000 to 5,000 estimate A small freshwater cnidarian. It's controlled by a diffuse nerve net and has no brain. * [Box jellyfish](https://simulation-hypothesis.com/entities/jellyfish)Tripedalia cystophora No published count A box jellyfish has no central brain, yet it showed associative learning. ### Invertebrates with brains * [Roundworm](https://simulation-hypothesis.com/entities/celegans)Caenorhabditis elegans 302 measured A worm about a millimetre long with exactly 302 neurons. It was the first animal to have its complete wiring diagram mapped. * [Fruit fly](https://simulation-hypothesis.com/entities/drosophila)Drosophila melanogaster 139,255 measured The largest complete brain wiring diagram mapped so far, covering about 139,000 neurons. * [Honeybee](https://simulation-hypothesis.com/entities/honeybee)Apis mellifera about 1 million estimate About a million neurons, and bees can learn abstract concepts such as same and different. * [Ant (Camponotus / Formica)](https://simulation-hypothesis.com/entities/ant)Camponotus spp. No published count A small brain that handles long-distance navigation by counting steps (path integration) and remembering landmarks. * [Portia jumping spider](https://simulation-hypothesis.com/entities/portia)Portia fimbriata / labiata No published count A jumping spider that plans detours. It takes routes that lead away from its prey and out of sight. * [Shore crab / crayfish](https://simulation-hypothesis.com/entities/decapod)Carcinus maenas; Procambarus / Astacus spp. No published count Crabs and crayfish. A 2021 government review found strong evidence that true crabs are sentient. * [Common octopus](https://simulation-hypothesis.com/entities/octopus)Octopus vulgaris 500 million estimate Most of its neurons are in its arms. A 2021 review found very strong evidence that it's sentient. ### Fish and amphibians * [Zebrafish](https://simulation-hypothesis.com/entities/zebrafish)Danio rerio No published count A small fish widely used in neuroscience. Its neurons haven't been fully counted. The adult brain has about 10 million cells. * [Cleaner wrasse](https://simulation-hypothesis.com/entities/wrasse)Labroides dimidiatus No published count A reef fish that passed a version of the mirror mark test. People disagree about what that shows. * [Frog](https://simulation-hypothesis.com/entities/frog)Anura (e.g. Xenopus / Rana) No published count The 2024 New York Declaration says there's a realistic possibility of conscious experience in amphibians. ### Birds * [Rock pigeon](https://simulation-hypothesis.com/entities/pigeon)Columba livia 310 million measured Hundreds of millions of neurons in a small brain. Pigeons can sort complex images into categories. * [Common raven](https://simulation-hypothesis.com/entities/raven)Corvus corax 2.17 billion measured Over two billion neurons, densely packed. Ravens remember individual competitors. * [New Caledonian crow](https://simulation-hypothesis.com/entities/nccrow)Corvus moneduloides No published count It makes hooked tools from twigs and wire. Its neurons haven't been counted. * [African grey parrot](https://simulation-hypothesis.com/entities/greyparrot)Psittacus erithacus 1.57 billion measured About 1.6 billion neurons. Decades of laboratory study of numerosity and categorisation. ### Mammals * [Mouse](https://simulation-hypothesis.com/entities/mouse)Mus musculus 70.9 million measured About 71 million neurons. Mice show facial expressions of pain that can be measured. * [Rat](https://simulation-hypothesis.com/entities/rat)Rattus norvegicus 200 million measured About 200 million neurons. Rats release trapped cagemates. * [Domestic cat](https://simulation-hypothesis.com/entities/cat)Felis catus 1.22 billion measured About 250 million neurons in the cerebral cortex. * [Domestic dog](https://simulation-hypothesis.com/entities/dog)Canis familiaris 1.8 to 2.6 billion, depending on the dog measured Up to about 600 million cortical neurons, depending on the dog. That's more than cats. * [Domestic pig](https://simulation-hypothesis.com/entities/pig)Sus scrofa domesticus 2.22 billion measured Pigs learn what a mirror shows and use it to find food. * [Bottlenose dolphin](https://simulation-hypothesis.com/entities/dolphin)Tursiops truncatus No published count Passes mirror self-recognition tests. There's no measured neuron count for the bottlenose dolphin. * [African elephant](https://simulation-hypothesis.com/entities/elephant)Loxodonta africana 257 billion measured 257 billion neurons, three times as many as a human, almost all in the cerebellum. * [Chimpanzee](https://simulation-hypothesis.com/entities/chimp)Pan troglodytes about 28 billion (predicted from scaling rules, not counted) estimate One of our two closest living relatives (along with the bonobo). It passes the mirror test. * [Human](https://simulation-hypothesis.com/entities/human)Homo sapiens 86.1 billion measured 86 billion neurons, 16 billion of them in the cerebral cortex. We're the only species whose members can give verbal reports of their experience. ## Collective systems Groups that make decisions no single member makes alone. They're studied as collective intelligence, not as minds. * [Ant colonies](https://simulation-hypothesis.com/entities/c-ants)Colonies pick the better nest because individual ants follow a quorum threshold. No ant compares the sites overall. * [Honeybee swarms](https://simulation-hypothesis.com/entities/c-bees)Swarms pick the best cavity through competition between scouts and a quorum. Stop signals provide cross-inhibition, which breaks deadlocks. * [Starling flocks](https://simulation-hypothesis.com/entities/c-starlings)Each bird tracks a fixed number of neighbours (about 6-7), not every bird within a fixed distance. This is called topological, not metric, interaction. * [Fish shoals](https://simulation-hypothesis.com/entities/c-fish)Shoals use quorum rules to decide. Large groups can sense environmental gradients that no single fish can. ## Artificial systems Kinds of machine intelligence, sorted by how they work. No current system is considered a strong candidate for consciousness. * [Rule-based expert systems](https://simulation-hypothesis.com/entities/ai-expert)Programs that encode a domain expert's knowledge as explicit if-then rules. An inference engine applies the rules in sequence to reach conclusions, usually with an explanation of its reasoning. * [Game-tree search engines (chess)](https://simulation-hypothesis.com/entities/ai-chess)Programs that enumerate possible move sequences to a fixed depth and score the resulting positions with a hand-built evaluation function. They prune branches that can't change the choice (minimax with alpha-beta pruning). * [Deep reinforcement learning agents](https://simulation-hypothesis.com/entities/ai-drl)Neural networks trained by trial and error to maximise a reward signal. They learn a policy or value function by interacting with an environment, often a simulated one such as a game. * [Large language models](https://simulation-hypothesis.com/entities/ai-llm)Transformer neural networks trained to predict the next token in very large text corpora, then often tuned on human feedback. * [Embodied robots](https://simulation-hypothesis.com/entities/ai-robots)Machines with sensors and actuators that act in the physical world. They often use layered reactive control instead of a central world model, so behaviour emerges from the loop between body and environment. * [Neuromorphic hardware](https://simulation-hypothesis.com/entities/ai-neuromorphic)Chips whose circuits imitate spiking neurons and synapses. They are event-driven and use little power, unlike a conventional processor-and-memory design. ## Hypothetical minds Minds proposed in the scientific and philosophical literature. Nobody has observed one. * [Whole-brain emulation](https://simulation-hypothesis.com/entities/h-wbe)unobserved Scanning a brain in fine detail and running a software model of it. * [Simulated minds in ancestor simulations](https://simulation-hypothesis.com/entities/h-ancestor)unobserved Bostrom's trilemma: either almost no civilisations reach a posthuman stage, or almost none run ancestor simulations, or we almost certainly live in one. * [Boltzmann brains](https://simulation-hypothesis.com/entities/h-boltzmann)unobserved Brain-like observers that appear out of rare thermal or quantum fluctuations. * [Mind uploading via gradual replacement](https://simulation-hypothesis.com/entities/h-uploading)unobserved Replacing neurons one by one with functionally identical artificial parts. * [Superintelligence](https://simulation-hypothesis.com/entities/h-superintelligence)unobserved A hypothetical intellect greatly exceeding human performance in virtually all domains. Bostrom's book analyses the routes to it (whole-brain emulation, biological cognition, AI), its forms and the control problems. Data version 2.0.0\. [Download (JSON)](https://simulation-hypothesis.com/entities/minds.json).