Computacao de treino, parametros, tamanho dos conjuntos de dados, hardware e acessibilidade dos sistemas que definiram a area, dos anos 1950 ate hoje.
| Grok 4 | xAI | Jul 9, 2025 | 5 x 10^26 FLOP | 3T | API access |
| GPT-4.5 | OpenAI | Feb 27, 2025 | 3.8 x 10^26 FLOP | — | API access |
| Grok 3 | xAI | Feb 17, 2025 | 3.5 x 10^26 FLOP | 3T | API access |
| GPT-5 | OpenAI | Aug 7, 2025 | 6.6 x 10^25 FLOP | — | API access |
| Llama 4 Behemoth (preview) | Meta AI | Apr 5, 2025 | 5.2 x 10^25 FLOP | 2T | Unreleased |
| Gemini 1.0 Ultra | Google DeepMind | Dec 6, 2023 | 5 x 10^25 FLOP | — | API access |
| Llama Nemotron Ultra 253B | NVIDIA | Mar 18, 2025 | 3.9 x 10^25 FLOP | 253B | Open weights (restricted use) |
| Composer 2.5 | Cursor | May 18, 2026 | 3.9 x 10^25 FLOP | 1T | — |
| Llama 3.1-405B | Meta AI | Jul 23, 2024 | 3.8 x 10^25 FLOP | 405B | Open weights (restricted use) |
| Claude 3.7 Sonnet | Anthropic | Feb 24, 2025 | 3.4 x 10^25 FLOP | — | API access |
| Grok-2 | xAI | Aug 13, 2024 | 3 x 10^25 FLOP | — | API access |
| Claude 3.5 Sonnet | Anthropic | Jun 20, 2024 | 2.7 x 10^25 FLOP | — | API access |
| Doubao-pro | ByteDance | Oct 28, 2024 | 2.5 x 10^25 FLOP | 500B | API access |
| Composer 2 | Cursor | Mar 19, 2026 | 2.3 x 10^25 FLOP | 1T | API access |
| GPT-4 (Jun 2023) | OpenAI | Jun 13, 2023 | 2.1 x 10^25 FLOP | 1.8T | API access |
| GPT-4 (Mar 2023) | OpenAI | Mar 15, 2023 | 2.1 x 10^25 FLOP | 1.8T | API access |
| Kimi K3 | Moonshot | Jul 16, 2026 | 2 x 10^25 FLOP | 2.8T | Open weights (non-commercial) |
| Nemotron-4 340B | NVIDIA | Jun 14, 2024 | 1.8 x 10^25 FLOP | 340B | Open weights (unrestricted) |
| Qwen3-Max | Alibaba | Sep 5, 2025 | 1.5 x 10^25 FLOP | 1T | API access |
| Pangu Ultra | Huawei | Apr 10, 2025 | 1.1 x 10^25 FLOP | 135B | Hosted access (no API) |
| Aramco Metabrain AI | Saudi Aramco | Mar 4, 2024 | 10^25 FLOP | 250B | Unreleased |
| Inflection-2 | Inflection AI | Nov 22, 2023 | 10^25 FLOP | — | Hosted access (no API) |
| Emu3.5 | Beijing Academy of Artificial Intelligence / BAAI | Oct 30, 2025 | 9.9 x 10^24 FLOP | 34.1B | Open weights (unrestricted) |
| DeepSeek-V4-Pro | DeepSeek | Apr 24, 2026 | 9.7 x 10^24 FLOP | 1.6T | Open weights (unrestricted) |
| Qwen2.5-VL-72B | Alibaba | Feb 19, 2025 | 9.6 x 10^24 FLOP | 72B | Open weights (restricted use) |
| Grok-1.5 | xAI | Mar 28, 2024 | 9.3 x 10^24 FLOP | — | Hosted access (no API) |
| Reka Core | Reka AI | Apr 15, 2024 | 8.4 x 10^24 FLOP | 67B | API access |
| SEA-LION V3 Llama3.1 70B | AI Singapore | Dec 19, 2024 | 8 x 10^24 FLOP | 70B | Open weights (restricted use) |
| Inflection-2.5 | Inflection AI | Mar 7, 2024 | 8 x 10^24 FLOP | — | Hosted access (no API) |
| Llama 3.1-70B | Meta AI | Jul 23, 2024 | 7.9 x 10^24 FLOP | 70B | Open weights (restricted use) |
| Llama-3.1-Nemotron-70B-Instruct | NVIDIA, Meta AI | Jun 12, 2024 | 7.9 x 10^24 FLOP | — | Open weights (restricted use) |
| Llama 3-70B | Meta AI | Apr 18, 2024 | 7.9 x 10^24 FLOP | 70B | Open weights (restricted use) |
| Qwen2.5 Instruct (72B) | Alibaba | Sep 19, 2024 | 7.9 x 10^24 FLOP | 72.7B | Open weights (restricted use) |
| Qwen2.5-72B | Alibaba | Sep 19, 2024 | 7.8 x 10^24 FLOP | 72.7B | Open weights (unrestricted) |
| AMIE (Articulate Medical Intelligence Explorer) | Google DeepMind, Google Research | Apr 9, 2025 | 7.3 x 10^24 FLOP | 340B | Unreleased |
| PaLM 2 | May 10, 2023 | 7.3 x 10^24 FLOP | 340B | API access | |
| Qwen3-32B | Alibaba | Apr 29, 2025 | 7.1 x 10^24 FLOP | 32.8B | Open weights (unrestricted) |
| MAI-Thinking-1 | Microsoft | Jun 2, 2026 | 7 x 10^24 FLOP | 962B | API access |
| Telechat2-115B | China Telecom | Sep 20, 2024 | 6.9 x 10^24 FLOP | 115B | Open weights (restricted use) |
| Llama 3.3 70B | Meta AI | Dec 6, 2024 | 6.9 x 10^24 FLOP | 70B | Open weights (restricted use) |
| GLM-5 | Z.ai (Zhipu AI) | Feb 17, 2026 | 6.8 x 10^24 FLOP | 744B | Open weights (unrestricted) |
| MiMo-V2.5-Pro | Xiaomi Corp | Apr 23, 2026 | 6.8 x 10^24 FLOP | 1T | Open weights (unrestricted) |
| Apertus 70B | ETH Zurich, Ecole Polytechnique Federale de Lausanne (EPFL), Swiss National Supercomputing Centre (CSCS), Swisscom | Sep 2, 2025 | 6.7 x 10^24 FLOP | 70B | Open weights (unrestricted) |
| Nemotron-H 56B | NVIDIA | Apr 14, 2025 | 6.7 x 10^24 FLOP | 56B | Open weights (non-commercial) |
| Nemotron 3 Ultra | NVIDIA | Jun 4, 2026 | 6.6 x 10^24 FLOP | 550B | Open weights (unrestricted) |
| Amazon Nova Pro | Amazon | Dec 3, 2024 | 6 x 10^24 FLOP | — | API access |
| Ling-1T | Ant Group | Oct 10, 2025 | 6 x 10^24 FLOP | 1T | Open weights (unrestricted) |
| Kimi K2.5 | Moonshot | Feb 2, 2026 | 5.8 x 10^24 FLOP | 1T | Open weights (unrestricted) |
| Hunyuan-TurboS | Tencent | Mar 11, 2025 | 5.4 x 10^24 FLOP | 560B | API access |
| gpt-oss-120b | OpenAI | Aug 5, 2025 | 4.9 x 10^24 FLOP | 116.8B | Open weights (unrestricted) |
| Amazon Titan | Amazon | Sep 28, 2023 | 4.8 x 10^24 FLOP | 200B | API access |
| Qwen3-235B-A22B (Jul 2025) | Alibaba | Jul 25, 2025 | 4.8 x 10^24 FLOP | 235B | Open weights (unrestricted) |
| Qwen3-235B-A22B-Thinking (Jul 2025) | Alibaba | Jul 25, 2025 | 4.8 x 10^24 FLOP | 235B | Open weights (unrestricted) |
| Qwen3-235B-A22B | Alibaba | Apr 29, 2025 | 4.8 x 10^24 FLOP | 235B | Open weights (unrestricted) |
| GLM-4.7 | Z.ai (Zhipu AI) | Dec 22, 2025 | 4.4 x 10^24 FLOP | 358B | Open weights (unrestricted) |
| GLM-4.6 | Z.ai (Zhipu AI), Tsinghua University | Sep 30, 2025 | 4.4 x 10^24 FLOP | 357B | Open weights (unrestricted) |
| GLM-4.5 | Z.ai (Zhipu AI), Tsinghua University | Aug 5, 2025 | 4.4 x 10^24 FLOP | 355B | Open weights (unrestricted) |
| MiniMax-M1-80k | MiniMax | Jun 13, 2025 | 4.3 x 10^24 FLOP | 456B | Open weights (unrestricted) |
| MiMo-V2.5 | Xiaomi Corp | Apr 23, 2026 | 4.3 x 10^24 FLOP | 310B | Open weights (unrestricted) |
| AFM-server | Apple | Jul 29, 2024 | 4.3 x 10^24 FLOP | — | Hosted access (no API) |
| DeepSeek-V3.2 | DeepSeek | Dec 1, 2025 | 4.2 x 10^24 FLOP | — | Open weights (unrestricted) |
| Kimi K2 Thinking | Moonshot | Nov 6, 2025 | 4.2 x 10^24 FLOP | 1T | Open weights (restricted use) |
| MiniMax-M1-40k | MiniMax | Jun 13, 2025 | 4.2 x 10^24 FLOP | 456B | Open weights (unrestricted) |
| DeepSeek-V3.2-Exp | DeepSeek | Sep 29, 2025 | 4.2 x 10^24 FLOP | 671B | Open weights (unrestricted) |
| Step-Video-T2V | StepFun | Feb 24, 2025 | 4.1 x 10^24 FLOP | 30B | Open weights (unrestricted) |
| Llama 4 Scout | Meta AI | Apr 5, 2025 | 4.1 x 10^24 FLOP | 109B | Open weights (restricted use) |
| DeepSeek-R1 (May 2025) | DeepSeek | May 28, 2025 | 4 x 10^24 FLOP | 671B | Open weights (unrestricted) |
| MegaScale (Production) | ByteDance, Peking University | Feb 23, 2024 | 3.9 x 10^24 FLOP | 530B | Unreleased |
| SenseChat | SenseTime | Apr 10, 2023 | 3.9 x 10^24 FLOP | 180B | API access |
| Claude 2 | Anthropic | Jul 11, 2023 | 3.9 x 10^24 FLOP | — | API access |
| EXAONE 4.5 | LG AI Research | Apr 9, 2026 | 3.9 x 10^24 FLOP | 33.2B | Open weights (restricted use) |
| Falcon-180B | Technology Innovation Institute | Sep 6, 2023 | 3.8 x 10^24 FLOP | 180B | Open weights (restricted use) |
| LongCat-Flash | Meituan Inc | Sep 1, 2025 | 3.7 x 10^24 FLOP | 560B | Open weights (unrestricted) |
| Falcon-H1 | Technology Innovation Institute | May 21, 2025 | 3.7 x 10^24 FLOP | 34B | Open weights (unrestricted) |
| Cosmos-Transfer1-7B | NVIDIA | Mar 18, 2025 | 3.6 x 10^24 FLOP | 7B | Open weights (restricted use) |
| DeepSeek-V3.1-Terminus | DeepSeek | Sep 22, 2025 | 3.6 x 10^24 FLOP | 671B | Open weights (unrestricted) |
| DeepSeek-V3.1 | DeepSeek | Aug 21, 2025 | 3.6 x 10^24 FLOP | 671B | Open weights (unrestricted) |
| QwQ-32B | Alibaba | Mar 6, 2025 | 3.5 x 10^24 FLOP | 32.5B | Open weights (unrestricted) |
| Qwen2.5-32B | Alibaba | Sep 17, 2024 | 3.5 x 10^24 FLOP | 32.5B | Open weights (unrestricted) |
| DeepSeek-R1 | DeepSeek | Jan 20, 2025 | 3.5 x 10^24 FLOP | 671B | Open weights (unrestricted) |
| Hunyuan-Large | Tencent | Nov 6, 2024 | 3.5 x 10^24 FLOP | 389B | Open weights (restricted use) |
| DeepSeek-V3 (Mar 2025) | DeepSeek | Mar 24, 2025 | 3.3 x 10^24 FLOP | 671B | Open weights (restricted use) |
| DeepSeek-V3 | DeepSeek | Dec 24, 2024 | 3.3 x 10^24 FLOP | 671B | Open weights (restricted use) |
| Qwen3-14B | Alibaba | Apr 29, 2025 | 3.2 x 10^24 FLOP | 14.8B | Open weights (unrestricted) |
| MiniMax-Text-01 | MiniMax | Jan 14, 2025 | 3.1 x 10^24 FLOP | 456B | Open weights (restricted use) |
| Pangu Ultra MoE | Huawei | May 7, 2025 | 3.1 x 10^24 FLOP | 718B | Hosted access (no API) |
| NVLM-X 72B | NVIDIA | Oct 22, 2024 | 3 x 10^24 FLOP | 72B | Unreleased |
| LLaVA-OV-72B | ByteDance, Nanyang Technological University, Chinese University of Hong Kong (CUHK), Hong Kong University of Science and Technology (HKUST) | Aug 6, 2024 | 3 x 10^24 FLOP | 72B | Open weights (unrestricted) |
| NVLM-D 72B | NVIDIA | Oct 22, 2024 | 3 x 10^24 FLOP | 72B | Open weights (non-commercial) |
| NVLM-H 72B | NVIDIA | Oct 22, 2024 | 3 x 10^24 FLOP | 72B | Unreleased |
| Qwen2-72B | Alibaba | Jun 7, 2024 | 3 x 10^24 FLOP | 72.7B | Open weights (unrestricted) |
| Kimi K2 | Moonshot | Jul 11, 2025 | 3 x 10^24 FLOP | 1T | Open weights (restricted use) |
| Grok-1 | xAI | Nov 4, 2023 | 2.9 x 10^24 FLOP | 314B | Open weights (unrestricted) |
| GLM-4-32B-0414 | Z.ai (Zhipu AI), Tsinghua University | Apr 14, 2025 | 2.9 x 10^24 FLOP | 32B | Open weights (unrestricted) |
| GLM-Z1-Rumination-32B-0414 | Tsinghua University | Apr 14, 2025 | 2.9 x 10^24 FLOP | 32B | Open weights (unrestricted) |
| ERNIE-4.5-300B-A47B | Baidu | Jun 29, 2025 | 2.8 x 10^24 FLOP | 300B | Open weights (unrestricted) |
| Cosmos-Predict1-14b-Video2World | NVIDIA | Jan 7, 2025 | 2.8 x 10^24 FLOP | 14B | Open weights (restricted use) |
| Cosmos-1.0- Diffusion-14B Video2World | NVIDIA | Jan 7, 2025 | 2.8 x 10^24 FLOP | 14B | Open weights (restricted use) |
| Minerva (540B) | Jun 29, 2022 | 2.7 x 10^24 FLOP | 540.4B | Unreleased | |
| EXAONE 4.0 (32B) | LG AI Research | Jul 15, 2025 | 2.7 x 10^24 FLOP | 32B | Open weights (non-commercial) |
| DBRX | Databricks | Mar 27, 2024 | 2.6 x 10^24 FLOP | 132B | Open weights (restricted use) |
| Seed-OSS-36B-Base | ByteDance | Aug 21, 2025 | 2.6 x 10^24 FLOP | — | Open weights (unrestricted) |
| GPT-3.5 (davinci-002) | OpenAI | Mar 15, 2022 | 2.6 x 10^24 FLOP | — | API access |
| Step-Omni | StepFun | Feb 18, 2025 | 2.5 x 10^24 FLOP | 130B | Unreleased |
| Flan-PaLM 540B | Oct 20, 2022 | 2.5 x 10^24 FLOP | 540B | Unreleased | |
| U-PaLM (540B) | Oct 20, 2022 | 2.5 x 10^24 FLOP | 540B | Unreleased | |
| PaLM (540B) | Google Research | Apr 4, 2022 | 2.5 x 10^24 FLOP | 540.4B | Unreleased |
| DeepSeek-V4-Flash | DeepSeek | Apr 24, 2026 | 2.5 x 10^24 FLOP | 284B | Open weights (unrestricted) |
| MiMo-V2-Flash | Xiaomi Corp | Dec 16, 2025 | 2.4 x 10^24 FLOP | 309B | Open weights (unrestricted) |
| Gemma-SEA-LION-v4-27B-IT | AI Singapore | Aug 20, 2025 | 2.3 x 10^24 FLOP | 27B | Open weights (restricted use) |
| Mixtral 8x22B | Mistral AI | Apr 17, 2024 | 2.3 x 10^24 FLOP | 141B | Open weights (unrestricted) |
| Gemma 3 27B | Google DeepMind | Mar 12, 2025 | 2.3 x 10^24 FLOP | 27B | Open weights (restricted use) |
| Evo 2 40B | Arc Institute, Stanford University, NVIDIA, Liquid, University of California (UC) Berkeley, Goodfire, Columbia University, University of California San Francisco | Feb 19, 2025 | 2.3 x 10^24 FLOP | 40.3B | Open weights (unrestricted) |
| Llama 4 Maverick | Meta AI | Apr 5, 2025 | 2.2 x 10^24 FLOP | 400B | Open weights (restricted use) |
| MiniMax-VL-01 | MiniMax | Jan 14, 2025 | 2.1 x 10^24 FLOP | — | Open weights (restricted use) |
| TxGemma 27B | Google DeepMind, Google Research | Apr 8, 2025 | 2.1 x 10^24 FLOP | 27B | Open weights (restricted use) |
| Gemma 2 27B | Google DeepMind | Jun 24, 2024 | 2.1 x 10^24 FLOP | 27B | Open weights (restricted use) |
| FLAN 137B | Google Research | Sep 3, 2021 | 2 x 10^24 FLOP | 137B | Unreleased |
| Step-Audio-Chat 130B | StepFun | Feb 18, 2025 | 1.9 x 10^24 FLOP | 130B | Open weights (unrestricted) |
| Inkling | Thinking Machines | Jul 15, 2026 | 1.8 x 10^24 FLOP | 975B | Open weights (unrestricted) |
| GLM-4.5V | Z.ai (Zhipu AI), Tsinghua University | Aug 15, 2025 | 1.8 x 10^24 FLOP | 108B | Open weights (unrestricted) |
| Yi-Large | 01.AI | May 13, 2024 | 1.8 x 10^24 FLOP | 100B | API access |
| DeepSeek-V2.5 | DeepSeek | Sep 6, 2024 | 1.8 x 10^24 FLOP | 236B | Open weights (restricted use) |
| Qwen3-8B | Alibaba | Apr 29, 2025 | 1.8 x 10^24 FLOP | 8.2B | Open weights (unrestricted) |
| EXAONE 1.0 | LG | Dec 14, 2021 | 1.7 x 10^24 FLOP | 300B | Unreleased |
| GLM-4.5-Air | Z.ai (Zhipu AI), Tsinghua University | Aug 5, 2025 | 1.7 x 10^24 FLOP | 106B | Open weights (unrestricted) |
| Movie Gen Video | Meta AI | Oct 4, 2024 | 1.6 x 10^24 FLOP | 30B | Unreleased |
| Chameleon-34B | Facebook AI Research | May 16, 2024 | 1.6 x 10^24 FLOP | 34B | Open weights (non-commercial) |
| Qwen2.5-14B | Alibaba | Sep 19, 2024 | 1.6 x 10^24 FLOP | 14.7B | Open weights (unrestricted) |
| Qwen3-Coder-480B-A35B | Alibaba | Jul 22, 2025 | 1.6 x 10^24 FLOP | 480B | Open weights (unrestricted) |
| Ling-Plus ("Bailing") | Ant Group | Mar 10, 2025 | 1.6 x 10^24 FLOP | 290B | Open weights (unrestricted) |
| Wu Dao 2.0 | Beijing Academy of Artificial Intelligence / BAAI | May 31, 2021 | 1.5 x 10^24 FLOP | 1.8T | API access |
| NVIDIA-Nemotron-Nano-9B-v2 | NVIDIA | Aug 18, 2025 | 1.5 x 10^24 FLOP | 9B | Open weights (restricted use) |
| K-EXAONE | LG AI Research | Dec 31, 2025 | 1.5 x 10^24 FLOP | 236B | API access |
| NVIDIA-Nemotron-Nano-12B-v2 | NVIDIA | Aug 18, 2025 | 1.5 x 10^24 FLOP | 12B | Open weights (restricted use) |
| Yi-Lightning | 01.AI | Oct 18, 2024 | 1.5 x 10^24 FLOP | — | API access |
| Solar Open 100B | Upstage | Dec 31, 2025 | 1.4 x 10^24 FLOP | 102B | API access |
| Cosmos-Predict1-7b-Video2World | NVIDIA | Jan 7, 2025 | 1.4 x 10^24 FLOP | 7B | Open weights (restricted use) |
| Seed1.5-VL | ByteDance | May 11, 2025 | 1.4 x 10^24 FLOP | — | API access |
| Arcee Trinity Large | Prime Intellect, Arcee AI | Feb 19, 2026 | 1.3 x 10^24 FLOP | 398B | Open weights (restricted use) |
| OLMo 2 32B | Allen Institute for AI | Mar 13, 2025 | 1.3 x 10^24 FLOP | 32B | Open weights (unrestricted) |
| Qwen1.5-72B | Alibaba | Feb 4, 2024 | 1.3 x 10^24 FLOP | 72B | Open weights (restricted use) |
| Qwen-72B | Alibaba | Nov 30, 2023 | 1.3 x 10^24 FLOP | 72B | Open weights (restricted use) |
| Pangu Pro MoE | Huawei | May 28, 2025 | 1.3 x 10^24 FLOP | 72B | Open weights (unrestricted) |
| DeepSeek-Coder-V2 236B | DeepSeek | Jun 17, 2024 | 1.3 x 10^24 FLOP | 236B | Open weights (restricted use) |
| EXAONE 3.5-R 32B | LG AI Research | Mar 14, 2025 | 1.3 x 10^24 FLOP | 32B | Unreleased |
| EXAONE Deep 32B | LG AI Research | Mar 16, 2025 | 1.3 x 10^24 FLOP | 32B | Open weights (non-commercial) |
| Code Llama-70B | Meta AI | Jan 29, 2024 | 1.3 x 10^24 FLOP | 70B | Open weights (restricted use) |
| EXAONE 3.5 32B | LG AI Research | Dec 9, 2024 | 1.3 x 10^24 FLOP | 32B | Open weights (non-commercial) |
| XVERSE-65B-2 | XVERSE Technology, Shenzhen Yuanxiang Technology | Dec 8, 2023 | 1.2 x 10^24 FLOP | 65B | Open weights (restricted use) |
| SEA-LION V3 Llama3.1 8B | AI Singapore | Dec 19, 2024 | 1.2 x 10^24 FLOP | 8B | Open weights (restricted use) |
| Foundation-sec-8b | Cisco | Apr 28, 2025 | 1.2 x 10^24 FLOP | 8B | Open weights (unrestricted) |
| Llama 3.1-8B | Meta AI | Jul 23, 2024 | 1.2 x 10^24 FLOP | 8B | Open weights (restricted use) |
| dots.llm1 | Rednote | Jul 6, 2025 | 1.2 x 10^24 FLOP | 142B | Open weights (unrestricted) |
| Granite-4.0-H-Small | IBM | Oct 2, 2025 | 1.2 x 10^24 FLOP | 32B | Open weights (unrestricted) |
| PLaMo-100B | Preferred Networks Inc | Jun 14, 2024 | 1.2 x 10^24 FLOP | 100B | API access |
| Mi:dm 200B | KT | Oct 31, 2023 | 1.2 x 10^24 FLOP | 200B | API access |
| Hunyuan | Tencent | Sep 7, 2023 | 1.2 x 10^24 FLOP | 100B | API access |
| Luca 2.0 | Mianbi Intelligence | Aug 29, 2023 | 1.2 x 10^24 FLOP | 100B | Hosted access (no API) |
| Mistral Small 3 | Mistral AI | Jan 30, 2025 | 1.2 x 10^24 FLOP | 24B | Open weights (unrestricted) |
| MiMo-VL-7B-SFT | Xiaomi Corp | Jun 4, 2025 | 1.2 x 10^24 FLOP | 7B | Open weights (unrestricted) |
| Olmo 3 | Allen Institute for AI | Nov 20, 2025 | 1.1 x 10^24 FLOP | 32B | Open weights (unrestricted) |
| Qwen2.5-Coder (32B) | Alibaba | Nov 12, 2024 | 1.1 x 10^24 FLOP | 32.5B | Open weights (unrestricted) |
| ESM3 (98B) | EvolutionaryScale, University of California (UC) Berkeley | Jun 25, 2024 | 1.1 x 10^24 FLOP | 98.5B | Unreleased |
| ALLaM adapted 70B | Saudi Data and Artificial Intelligence Authority | May 21, 2024 | 1.1 x 10^24 FLOP | 70B | Unreleased |
| ALLaM 34B | Saudi Data and Artificial Intelligence Authority | May 21, 2024 | 1.1 x 10^24 FLOP | 34B | Unreleased |
| Solar Open2 250B | Upstage | Jun 28, 2026 | 1.1 x 10^24 FLOP | 250.3B | Open weights (unrestricted) |
| MiMo-7B-Base | Xiaomi Corp | Jun 5, 2025 | 1.1 x 10^24 FLOP | 7B | Open weights (unrestricted) |
| BlueLM 175B | vivo AI lab | Nov 2, 2023 | 1.1 x 10^24 FLOP | 175B | Unreleased |
| ERNIE 3.0 Titan | Baidu, Peng Cheng Laboratory | Dec 23, 2021 | 10^24 FLOP | 260B | Hosted access (no API) |
| DeepSeek-V2 (MoE-236B) | DeepSeek | May 7, 2024 | 10^24 FLOP | 236B | Open weights (restricted use) |
| TigerBot-70B | Tigerobo | Sep 6, 2023 | 10^24 FLOP | 70B | Open weights (restricted use) |
| Inflection-1 | Inflection AI | Jun 23, 2023 | 10^24 FLOP | — | Hosted access (no API) |
| InternLM | Shanghai AI Lab, SenseTime | Jul 6, 2023 | 10 x 10^23 FLOP | 104B | Unreleased |
| Qwen2.5-VL-7B | Alibaba | Feb 19, 2025 | 9.9 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| MegaScale (530B) | ByteDance, Peking University | Feb 23, 2024 | 9.7 x 10^23 FLOP | 530B | Unreleased |
| Jais-70B | G42, Inception G42 | Aug 5, 2024 | 9.7 x 10^23 FLOP | 70B | Open weights (restricted use) |
| Phi-4-Reasoning | Microsoft | Apr 30, 2025 | 9.3 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| Phi-4 | Microsoft Research | Dec 12, 2024 | 9.3 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| GLM-4.1V-Thinking | Z.ai (Zhipu AI), Tsinghua University | Aug 15, 2025 | 9.2 x 10^23 FLOP | 9B | Open weights (unrestricted) |
| Seaweed-7B | ByteDance | Apr 11, 2025 | 9 x 10^23 FLOP | 7B | Unreleased |
| Lapa LLM | Ukrainian Catholic University, Igor Sikorsky Kyiv Polytechnic Institute, AGH University of Krakow, Lviv Polytechnic | Oct 25, 2025 | 8.6 x 10^23 FLOP | 12B | Open weights (unrestricted) |
| Qwen3-4B | Alibaba | Apr 29, 2025 | 8.6 x 10^23 FLOP | 4B | Open weights (unrestricted) |
| Gemma 3 12B | Google DeepMind | Mar 12, 2025 | 8.6 x 10^23 FLOP | 12B | Open weights (restricted use) |
| Qwen2.5-Math-7B-Base | Alibaba | Sep 19, 2024 | 8.6 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Megatron-Turing NLG 530B | Microsoft, NVIDIA | Oct 11, 2021 | 8.6 x 10^23 FLOP | 530B | Unreleased |
| Baichuan2-53B | Baichuan | Aug 9, 2023 | 8.3 x 10^23 FLOP | 53B | Unreleased |
| Qwen2.5-7B | Alibaba | Sep 19, 2024 | 8.2 x 10^23 FLOP | 7.6B | Open weights (unrestricted) |
| SambaLingo-Thai-Chat-70B | SambaNova Systems, Inc | Apr 8, 2024 | 8.1 x 10^23 FLOP | 70B | Open weights (restricted use) |
| GLM-4-9B-0414 | Tsinghua University | Apr 14, 2025 | 8.1 x 10^23 FLOP | 9B | Open weights (unrestricted) |
| Llama 2-70B | Meta AI | Jul 18, 2023 | 8.1 x 10^23 FLOP | 70B | Open weights (restricted use) |
| DeepSeek LLM 67B | DeepSeek | Jan 5, 2024 | 8 x 10^23 FLOP | 67B | Open weights (restricted use) |
| BlueLM 130B | vivo AI lab | Nov 2, 2023 | 7.8 x 10^23 FLOP | 130B | Unreleased |
| Mixtral 8x7B | Mistral AI | Dec 11, 2023 | 7.7 x 10^23 FLOP | 46.7B | Open weights (unrestricted) |
| Ring-flash-linear-2.0 | Ant Group | Oct 23, 2025 | 7.7 x 10^23 FLOP | 104.2B | Open weights (unrestricted) |
| Nemotron-4 15B | NVIDIA | Feb 27, 2024 | 7.5 x 10^23 FLOP | 15B | Unreleased |
| YandexGPT 5 Lite | Yandex | Feb 25, 2025 | 7.4 x 10^23 FLOP | 8B | Open weights (restricted use) |
| Yi-1.5-34B | 01.AI | May 13, 2024 | 7.3 x 10^23 FLOP | 34B | Open weights (restricted use) |
| Ling-flash-base-2.0-20T | Ant Group | Sep 10, 2025 | 7.3 x 10^23 FLOP | 100B | Open weights (unrestricted) |
| Llama SEA-LION V2 8B | AI Singapore | Jul 30, 2024 | 7.2 x 10^23 FLOP | 8B | Open weights (restricted use) |
| ProtChatGPT | University of Technology Sydney, Zhejiang University (ZJU) | Feb 15, 2024 | 7.2 x 10^23 FLOP | 8B | Unreleased |
| P-LLama3 | University of Siena | Aug 12, 2024 | 7.2 x 10^23 FLOP | 8B | Open weights (restricted use) |
| Nemotron-H 8B | NVIDIA | Apr 14, 2025 | 7.2 x 10^23 FLOP | 8B | Open weights (non-commercial) |
| Llama 3-8B | Meta AI | Apr 18, 2024 | 7.2 x 10^23 FLOP | 8B | Open weights (restricted use) |
| AgentFounder-30B | Alibaba | Sep 16, 2025 | 6.5 x 10^23 FLOP | 30B | Open weights (unrestricted) |
| Qwen3-30B-A3B | Alibaba | Apr 29, 2025 | 6.5 x 10^23 FLOP | 30B | Open weights (unrestricted) |
| Gopher (280B) | DeepMind | Dec 8, 2021 | 6.3 x 10^23 FLOP | 280B | Unreleased |
| Reka Flash | Reka AI | Apr 15, 2024 | 6.3 x 10^23 FLOP | 21B | API access |
| Step-1 | StepFun | Feb 18, 2025 | 6.2 x 10^23 FLOP | 130B | Unreleased |
| xTrimoPGLM -100B | Tsinghua University, BioMap Research | Jul 6, 2023 | 6.2 x 10^23 FLOP | 100B | Unreleased |
| Marin 8B | Marin | May 19, 2025 | 6.1 x 10^23 FLOP | 8B | Open weights (unrestricted) |
| Yi-34B | 01.AI | Nov 2, 2023 | 6.1 x 10^23 FLOP | 34B | Open weights (restricted use) |
| Qwen2-VL-72B | Alibaba | Sep 18, 2024 | 6 x 10^23 FLOP | 72B | Open weights (restricted use) |
| Falcon3-7B | Technology Innovation Institute | Dec 17, 2024 | 5.9 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Granite 3.1 8B | IBM | Dec 18, 2024 | 5.8 x 10^23 FLOP | 8.1B | Open weights (unrestricted) |
| Granite 3.0 8B | IBM | Oct 21, 2024 | 5.8 x 10^23 FLOP | 8.1B | Open weights (unrestricted) |
| Llama 3.2 11B | Meta AI | Sep 24, 2024 | 5.8 x 10^23 FLOP | 10.6B | Open weights (restricted use) |
| Chinchilla | DeepMind | Mar 29, 2022 | 5.8 x 10^23 FLOP | 70B | Unreleased |
| PaLI-X | Google Research | May 29, 2023 | 5.6 x 10^23 FLOP | 55B | Unreleased |
| BIG-G 137B | Jun 9, 2022 | 5.6 x 10^23 FLOP | 137B | Unreleased | |
| LIMA | Meta AI, Carnegie Mellon University (CMU), University of Southern California, Tel Aviv University | May 18, 2023 | 5.5 x 10^23 FLOP | 65B | Unreleased |
| Guanaco-65B | University of Washington | May 23, 2023 | 5.5 x 10^23 FLOP | 65B | Open weights (non-commercial) |
| LLaMA-65B | Meta AI | Feb 24, 2023 | 5.5 x 10^23 FLOP | 65.2B | Open weights (non-commercial) |
| gpt-oss-20b | OpenAI | Aug 5, 2025 | 5.5 x 10^23 FLOP | 20.9B | Open weights (unrestricted) |
| Llemma 34B | Princeton University, University of Toronto, Vector Institute, University of Cambridge, Carnegie Mellon University (CMU), University of Washington, EleutherAI | Oct 16, 2023 | 5.4 x 10^23 FLOP | 34B | Open weights (restricted use) |
| Code Llama-34B | Meta AI | Aug 14, 2023 | 5.3 x 10^23 FLOP | 34B | Open weights (restricted use) |
| Gemma 3n | May 20, 2025 | 5.2 x 10^23 FLOP | 7.8B | Open weights (restricted use) | |
| Parti | Google Research | Jun 22, 2022 | 5.1 x 10^23 FLOP | 20B | Unreleased |
| MM1-30B | Apple | Mar 14, 2024 | 4.9 x 10^23 FLOP | 30B | Unreleased |
| Nemotron 3-Nano-30B-A3B | NVIDIA | Dec 15, 2025 | 4.8 x 10^23 FLOP | 31.6B | Open weights (restricted use) |
| YaYi 2.0 | Yayi (Wenge) | Dec 22, 2023 | 4.8 x 10^23 FLOP | 30B | Open weights (restricted use) |
| PanGu-Σ | Huawei Noah's Ark Lab | Mar 20, 2023 | 4.7 x 10^23 FLOP | 1.1T | Unreleased |
| OLMo 2 Furious 13B | Allen Institute for AI, University of Washington, New York University (NYU) | Dec 31, 2024 | 4.6 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| AFM-on-device | Apple | Jul 29, 2024 | 4.5 x 10^23 FLOP | 2.7B | Hosted access (no API) |
| SEA-LION V3 Gemma2 9B | AI Singapore | Dec 19, 2024 | 4.5 x 10^23 FLOP | 9B | Open weights (restricted use) |
| Pharia-1-LLM-7B | Aleph Alpha | Aug 26, 2024 | 4.4 x 10^23 FLOP | 7B | Open weights (non-commercial) |
| MamayLM | INSAIT, ETH Zurich | Apr 23, 2025 | 4.4 x 10^23 FLOP | 9B | Open weights (unrestricted) |
| JIUTIAN-139MoE | China Mobile | Jun 15, 2024 | 4.4 x 10^23 FLOP | 38.8B | Open weights (unrestricted) |
| TxGemma 9B | Google DeepMind, Google Research | Apr 8, 2025 | 4.4 x 10^23 FLOP | 9B | Open weights (restricted use) |
| Gemma 2 9B | Google DeepMind | Jun 24, 2024 | 4.3 x 10^23 FLOP | 9B | Open weights (restricted use) |
| OPT-IML (175B) | Meta AI | Dec 22, 2022 | 4.3 x 10^23 FLOP | 175B | Open weights (non-commercial) |
| BlenderBot 3 | McGill University, Meta AI, Mila - Quebec AI (originally Montreal Institute for Learning Algorithms) | Aug 10, 2022 | 4.3 x 10^23 FLOP | 175B | Open weights (non-commercial) |
| OPT-175B | Meta AI | May 2, 2022 | 4.3 x 10^23 FLOP | 175B | Open weights (non-commercial) |
| EXAONE 3.5-R 7.8B | LG AI Research | Mar 14, 2025 | 4.3 x 10^23 FLOP | 7.8B | Unreleased |
| 360Zhinao2-7B | 360 Security Technology | Nov 18, 2024 | 4.2 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| EXAONE Deep 7.8B | LG AI Research | Mar 16, 2025 | 4.2 x 10^23 FLOP | 7.8B | Open weights (non-commercial) |
| EXAONE 3.5 7.8B | LG AI Research | Dec 9, 2024 | 4.2 x 10^23 FLOP | 7.8B | Open weights (non-commercial) |
| Wan 2.2 14B T2V | Alibaba | Jul 28, 2025 | 4.2 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| BlueLM 70B | vivo AI lab | Nov 2, 2023 | 4.2 x 10^23 FLOP | 70B | Unreleased |
| Llama 2-34B | Meta AI | Jul 18, 2023 | 4.1 x 10^23 FLOP | 34B | Unreleased |
| Qwen2.5-VL-3B | Alibaba | Feb 19, 2025 | 4.1 x 10^23 FLOP | 3B | Open weights (restricted use) |
| Nanbeige2-16B-Chat | Nanbeige LLM Lab | May 28, 2024 | 4 x 10^23 FLOP | 15.8B | Open weights (restricted use) |
| phi-3-medium 14B | Microsoft | Apr 23, 2024 | 4 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| EXAONE 3.0 | LG AI Research | Aug 7, 2024 | 4 x 10^23 FLOP | 7.8B | Open weights (non-commercial) |
| DeepSeek Coder 33B | DeepSeek, Peking University | Jan 25, 2024 | 4 x 10^23 FLOP | 33B | Open weights (restricted use) |
| Falcon Mamba | Technology Innovation Institute | Aug 12, 2024 | 3.9 x 10^23 FLOP | 7B | Open weights (restricted use) |
| StarCoder 2 15B | Hugging Face, ServiceNow, NVIDIA, BigCode | Feb 29, 2024 | 3.9 x 10^23 FLOP | 15B | Open weights (restricted use) |
| FunSearch | Google DeepMind | Dec 14, 2023 | 3.9 x 10^23 FLOP | 15B | Open weights (unrestricted) |
| Multi-Token Prediction 7B | Facebook AI Research | Apr 30, 2024 | 3.8 x 10^23 FLOP | 6.7B | Open weights (non-commercial) |
| Arctic | Snowflake | Apr 24, 2024 | 3.8 x 10^23 FLOP | 480B | Open weights (unrestricted) |
| Qwen2-57B-A14B | Alibaba | Jun 7, 2024 | 3.8 x 10^23 FLOP | 57B | Open weights (unrestricted) |
| Jurassic-1-Jumbo | AI21 Labs | Aug 11, 2021 | 3.7 x 10^23 FLOP | 178B | API access |
| Qwen3-1.7B | Alibaba | Apr 29, 2025 | 3.7 x 10^23 FLOP | 1.7B | Open weights (unrestricted) |
| BLOOM-176B | Hugging Face, BigScience | Jul 11, 2022 | 3.7 x 10^23 FLOP | 176.2B | Open weights (restricted use) |
| GLaM | Dec 13, 2021 | 3.6 x 10^23 FLOP | 1.2T | Unreleased | |
| Falcon 2 11B | Technology Innovation Institute | May 9, 2024 | 3.6 x 10^23 FLOP | 11B | Open weights (restricted use) |
| LaMDA | Feb 10, 2022 | 3.5 x 10^23 FLOP | 137B | Unreleased | |
| GLM-130B | Tsinghua University | Aug 4, 2022 | 3.5 x 10^23 FLOP | 130B | Open weights (non-commercial) |
| Luminous-supreme | Aleph Alpha | Aug 15, 2022 | 3.5 x 10^23 FLOP | 70B | API access |
| Yuan 1.0 | Inspur | Oct 12, 2021 | 3.5 x 10^23 FLOP | 245.7B | API access |
| OPT-13B | Meta AI | May 2, 2022 | 3.5 x 10^23 FLOP | 13B | — |
| DALL·E 2 | OpenAI | Apr 6, 2022 | 3.4 x 10^23 FLOP | 3.5B | API access |
| Qwen1.5-14B | Alibaba | Feb 4, 2024 | 3.4 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| FragLlama: Next-fragment prediction for molecular design | Facebook AI Research | May 16, 2024 | 3.3 x 10^23 FLOP | 7B | Open weights (non-commercial) |
| Qwen2.5-3B | Alibaba | Sep 19, 2024 | 3.3 x 10^23 FLOP | 3.1B | Open weights (non-commercial) |
| Galactica | Meta AI | Nov 16, 2022 | 3.2 x 10^23 FLOP | 120B | Open weights (non-commercial) |
| InstructGPT 175B | OpenAI | Jan 27, 2022 | 3.2 x 10^23 FLOP | 175B | API access |
| Granite-4.0-H-Micro | IBM | Oct 2, 2025 | 3.1 x 10^23 FLOP | 3B | Open weights (unrestricted) |
| GPT-3 175B (davinci) | OpenAI | May 28, 2020 | 3.1 x 10^23 FLOP | 174.6B | API access |
| InternLM2-20B | Shanghai AI Lab, SenseTime, Chinese University of Hong Kong (CUHK), Fudan University | Jan 12, 2024 | 3.1 x 10^23 FLOP | 20B | Open weights (restricted use) |
| CodeFuse-13B | Ant Group | Oct 10, 2023 | 3.1 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| Gemma 1.1 7B Instruct | Feb 24, 2024 | 3.1 x 10^23 FLOP | 8.5B | Open weights (restricted use) | |
| Gemma 7B | Google DeepMind | Feb 21, 2024 | 3.1 x 10^23 FLOP | 8.5B | Open weights (restricted use) |
| Phi-3.5-MoE | Microsoft | Apr 23, 2024 | 3 x 10^23 FLOP | 60.8B | Open weights (unrestricted) |
| Granite 20B | IBM Research | May 31, 2024 | 3 x 10^23 FLOP | 20B | Open weights (unrestricted) |
| Tri-21B | Trillion Labs | Aug 1, 2025 | 2.9 x 10^23 FLOP | 20.7B | Open weights (restricted use) |
| Qwen2-7B | Alibaba | Jun 7, 2024 | 2.9 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Stable LM 2 12B | Stability AI | Apr 8, 2024 | 2.9 x 10^23 FLOP | 12.1B | Open weights (restricted use) |
| ST-MoE | Google, Google Brain, Google Research | Feb 17, 2022 | 2.9 x 10^23 FLOP | 269B | Unreleased |
| MegaScale (175B) | ByteDance, Peking University | Feb 23, 2024 | 2.7 x 10^23 FLOP | 175B | Unreleased |
| LLaMA-33B | Meta AI | Feb 27, 2023 | 2.7 x 10^23 FLOP | 32.5B | Open weights (non-commercial) |
| Qwen3-Next-80B-A3B | Alibaba | Sep 10, 2025 | 2.7 x 10^23 FLOP | 80B | Open weights (unrestricted) |
| Whisper v3 | OpenAI | Nov 6, 2023 | 2.7 x 10^23 FLOP | 1.6B | Open weights (unrestricted) |
| Viking | Silo AI, University of Turku | Apr 4, 2024 | 2.6 x 10^23 FLOP | 33B | Open weights (unrestricted) |
| nekomata-14b | rinna | Dec 21, 2023 | 2.6 x 10^23 FLOP | 14.2B | Open weights (restricted use) |
| Qwen2.5-Coder (7B) | Alibaba | Sep 18, 2024 | 2.5 x 10^23 FLOP | 7.6B | Open weights (unrestricted) |
| Wan 2.1 14B I2V | Alibaba | Feb 25, 2025 | 2.5 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| Skywork-13B | Kunlun Inc. | Oct 30, 2023 | 2.5 x 10^23 FLOP | 13B | Open weights (restricted use) |
| Qwen-14B | Alibaba | Sep 28, 2023 | 2.5 x 10^23 FLOP | 14B | Open weights (restricted use) |
| Granite 13B | IBM | Nov 30, 2023 | 2.4 x 10^23 FLOP | 13B | API access |
| Nanbeige-16B | Nanbeige LLM Lab | Nov 1, 2023 | 2.4 x 10^23 FLOP | 16B | Open weights (unrestricted) |
| LightOn Mini | LightOn | Mar 21, 2023 | 2.4 x 10^23 FLOP | 40B | Hosted access (no API) |
| Falcon-40B | Technology Innovation Institute | Mar 15, 2023 | 2.4 x 10^23 FLOP | 40B | Open weights (unrestricted) |
| AlphaCode | DeepMind | Feb 2, 2022 | 2.4 x 10^23 FLOP | 41.1B | Unreleased |
| BloombergGPT | Bloomberg, Johns Hopkins University | Mar 30, 2023 | 2.4 x 10^23 FLOP | 50.6B | Unreleased |
| YaLM | Yandex | Jun 23, 2022 | 2.2 x 10^23 FLOP | 100B | Open weights (unrestricted) |
| Flamingo | DeepMind | Apr 29, 2022 | 2.2 x 10^23 FLOP | 80B | Unreleased |
| Teuken 7B | OpenGPT-X, Fraunhofer Institute for Algorithms and Scientific Computing, Forschungszentrum Julich, Technische Universität Dresden | Aug 21, 2025 | 2.1 x 10^23 FLOP | 7B | Open weights (non-commercial) |
| phi-3-small 7.4B | Microsoft | Apr 23, 2024 | 2.1 x 10^23 FLOP | 7.4B | Open weights (unrestricted) |
| Poro 34B | High-Performance Language Technologies (HPLT), University of Turku | Dec 14, 2023 | 2.1 x 10^23 FLOP | 34.2B | Open weights (unrestricted) |
| AlexaTM 20B | Amazon | Aug 2, 2022 | 2 x 10^23 FLOP | 19.8B | API access |
| Baichuan2-13B | Baichuan | Sep 6, 2023 | 2 x 10^23 FLOP | 13B | Open weights (restricted use) |
| HyperCLOVA 204B | NAVER | Sep 10, 2021 | 2 x 10^23 FLOP | 204B | Hosted access (no API) |
| SAM 3 | Meta AI | Nov 20, 2025 | 2 x 10^23 FLOP | 850M | Open weights (restricted use) |
| ViT-22B | Feb 10, 2023 | 1.9 x 10^23 FLOP | 21.7B | Unreleased | |
| Yi-1.5-9B | 01.AI | May 13, 2024 | 1.9 x 10^23 FLOP | 8.8B | Open weights (unrestricted) |
| MPT-30B | MosaicML | Jun 22, 2023 | 1.9 x 10^23 FLOP | 30B | Open weights (unrestricted) |
| MAP-Neo | University of Waterloo, 01.AI, Wuhan University | Jul 10, 2024 | 1.9 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Reka Edge | Reka AI | Apr 18, 2024 | 1.9 x 10^23 FLOP | 7B | API access |
| Mamba2-Hybrid | NVIDIA | Jun 12, 2024 | 1.8 x 10^23 FLOP | 8.7B | Open weights (unrestricted) |
| ERNIE-4.5-21B-A3B | Baidu | Jun 29, 2025 | 1.8 x 10^23 FLOP | 21B | Open weights (unrestricted) |
| OLMo 2 Furious 7B | Allen Institute for AI, University of Washington, New York University (NYU) | Dec 31, 2024 | 1.8 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Granite 3.1 2B | IBM | Dec 18, 2024 | 1.8 x 10^23 FLOP | 2.5B | Open weights (unrestricted) |
| Granite 3.0 2B | IBM | Oct 21, 2024 | 1.8 x 10^23 FLOP | 2.5B | Open weights (unrestricted) |
| Nemotron-3-8B | NVIDIA | Nov 15, 2023 | 1.8 x 10^23 FLOP | 8B | Open weights (restricted use) |
| Yuan 2.0 | Inspur | Nov 27, 2023 | 1.8 x 10^23 FLOP | 102.6B | Open weights (restricted use) |
| Sailor-7B-Chat | Sea AI Lab, Singapore University of Technology & Design | Apr 4, 2024 | 1.8 x 10^23 FLOP | 7.7B | Open weights (unrestricted) |
| Qwen2.5-Math-1.5B | Alibaba | Sep 19, 2024 | 1.8 x 10^23 FLOP | 1.5B | Open weights (unrestricted) |
| InternVL | Shanghai AI Lab, Nanjing University, The University of Hong Kong, Tsinghua University, SenseTime, University of Science and Technology of China (USTC) | Jan 15, 2024 | 1.7 x 10^23 FLOP | 14B | Open weights (unrestricted) |
| Llama 3.2 3B | Meta AI | Sep 24, 2024 | 1.7 x 10^23 FLOP | 3.2B | Open weights (restricted use) |
| Konan LLM 41B | Konan Technology | Dec 15, 2023 | 1.7 x 10^23 FLOP | 41B | Hosted access (no API) |
| ALLaM adapted13B | Saudi Data and Artificial Intelligence Authority | May 21, 2024 | 1.7 x 10^23 FLOP | 13B | API access |
| Ring-mini-linear-2.0 | Ant Group | Oct 23, 2025 | 1.7 x 10^23 FLOP | 16.4B | Open weights (unrestricted) |
| Ovis-7B | Alibaba, Nanjing University | Jun 17, 2024 | 1.7 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| PaLI | Sep 14, 2022 | 1.7 x 10^23 FLOP | 16.9B | Unreleased | |
| Ling-mini-base-2.0-20T | Ant Group | Sep 10, 2025 | 1.7 x 10^23 FLOP | 16B | Open weights (unrestricted) |
| Qwen1.5-7B | Alibaba | Feb 4, 2024 | 1.7 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Jiutian | China Mobile | Oct 12, 2023 | 1.7 x 10^23 FLOP | 13.9B | Open weights (unrestricted) |
| Qwen2.5-1.5B | Alibaba | Sep 19, 2024 | 1.7 x 10^23 FLOP | 1.5B | Open weights (unrestricted) |
| PLLaMa | University of California Santa Barbara (UCSB), University of Lincoln, Chinese Academy of Agricultural Sciences, Swedish University of Agricultural Sciences | Jan 3, 2024 | 1.6 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| Llama Guard | Meta AI | Dec 7, 2023 | 1.6 x 10^23 FLOP | 7B | Open weights (restricted use) |
| FinGPT-13B | University of California Los Angeles (UCLA), Columbia University, New York University (NYU) | Oct 7, 2023 | 1.6 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| Llama 2-13B | Meta AI | Jul 18, 2023 | 1.6 x 10^23 FLOP | 13B | Open weights (restricted use) |
| SparseOPT-175B | Institute of Science and Technology Austria (ISTA), Neural Magic | Jan 2, 2023 | 1.6 x 10^23 FLOP | 87.5B | Unreleased |
| StarCoder 2 7B | Hugging Face, ServiceNow, NVIDIA, BigCode | Feb 29, 2024 | 1.5 x 10^23 FLOP | 7B | Open weights (restricted use) |
| Multi-Token Prediction 13B | Facebook AI Research | Apr 30, 2024 | 1.5 x 10^23 FLOP | 13B | Unreleased |
| Ling-lite-1.5 ("Bailing") | Ant Group | Mar 10, 2025 | 1.5 x 10^23 FLOP | 16.8B | Open weights (unrestricted) |
| Hunyuan Video | Tencent | Dec 3, 2024 | 1.5 x 10^23 FLOP | 13B | Open weights (restricted use) |
| HyperCLOVA 82B | NAVER, Search Solutions | Sep 10, 2021 | 1.5 x 10^23 FLOP | 82B | API access |
| Aria | Rhymes AI | Oct 8, 2024 | 1.4 x 10^23 FLOP | 24.9B | Open weights (unrestricted) |
| 360Zhinao-7B | 360 Security Technology | May 22, 2024 | 1.4 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Uni-Med | Tsinghua University, Beijing University of Posts and Telecommunications | Nov 1, 2024 | 1.4 x 10^23 FLOP | 8.8B | Unreleased |
| Movie Gen Audio | Meta AI | Oct 4, 2024 | 1.4 x 10^23 FLOP | 13B | Unreleased |
| Granite-4.0-H-Tiny | IBM | Oct 2, 2025 | 1.3 x 10^23 FLOP | 7B | Open weights (unrestricted) |
| Qwen3-0.6B | Alibaba | Apr 29, 2025 | 1.3 x 10^23 FLOP | 600M | Open weights (unrestricted) |
| Yi 6B | 01.AI | Nov 2, 2023 | 1.3 x 10^23 FLOP | 6B | Open weights (restricted use) |
| VARCO LLM 2.0 base | NCSOFT | Aug 16, 2023 | 1.2 x 10^23 FLOP | 13B | API access |
| UL2 | Google Research, Google Brain | May 10, 2022 | 1.2 x 10^23 FLOP | 20B | Open weights (unrestricted) |
| Phi-4-Multimodal | Microsoft | Mar 3, 2025 | 1.2 x 10^23 FLOP | 5.6B | Open weights (unrestricted) |
| Llemma 7B | Princeton University, EleutherAI, University of Toronto, Vector Institute, University of Cambridge, Carnegie Mellon University (CMU), University of Washington | Oct 16, 2023 | 1.2 x 10^23 FLOP | 7B | Open weights (restricted use) |
| PLaMo-13B | Preferred Networks Inc | Sep 28, 2023 | 1.2 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| IDEFICS-80B | Hugging Face | Aug 22, 2023 | 1.2 x 10^23 FLOP | 80B | Open weights (non-commercial) |
| WizardCoder-15.5B | Microsoft | Jun 14, 2023 | 1.1 x 10^23 FLOP | 15.5B | Open weights (restricted use) |
| Meena | Google Brain | Jan 28, 2020 | 1.1 x 10^23 FLOP | 2.6B | Unreleased |
| OPT-66B | Meta AI | Jun 21, 2022 | 1.1 x 10^23 FLOP | 66B | Open weights (non-commercial) |
| Odyssey 102B | Anthrogen | Oct 18, 2025 | 1.1 x 10^23 FLOP | 102B | Unreleased |
| OpenVLA | Stanford University, University of California (UC) Berkeley, Toyota Research Institute, Google DeepMind, Massachusetts Institute of Technology (MIT), Physical Intelligence | Jun 13, 2024 | 1.1 x 10^23 FLOP | 7.2B | Open weights (unrestricted) |
| Code Llama-7B | Meta AI | Aug 14, 2023 | 1.1 x 10^23 FLOP | 7B | Open weights (restricted use) |
| Whisper v2 | OpenAI | Dec 5, 2022 | 1.1 x 10^23 FLOP | 1.6B | Open weights (unrestricted) |
| BlueLM 7B | vivo AI lab | Oct 31, 2023 | 1.1 x 10^23 FLOP | 7B | Open weights (restricted use) |
| Baichuan 2-7B | Baichuan | Sep 20, 2023 | 1.1 x 10^23 FLOP | 7B | Open weights (restricted use) |
| AlphaStar | DeepMind | Oct 30, 2019 | 1.1 x 10^23 FLOP | 139M | Unreleased |
| ruGPT-3.5 13B | Sber | Apr 24, 2023 | 1.1 x 10^23 FLOP | 13B | Open weights (unrestricted) |
| DCLM 7B | Apple | Jul 20, 2024 | 1.1 x 10^23 FLOP | 7B | Open weights (restricted use) |
| Jais-30b (phase 1) | Cerebras Systems, Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), Inception G42, G42 | Nov 8, 2023 | 10^23 FLOP | 30B | Open weights (unrestricted) |
| OLMo-7B | Allen Institute for AI, University of Washington | Feb 1, 2024 | 10^23 FLOP | 7B | Open weights (unrestricted) |
| DeepSeek-VL-7B | DeepSeek | Mar 8, 2024 | 10^23 FLOP | 7B | Open weights (restricted use) |
| Kimi Linear | Moonshot | Oct 30, 2025 | 10^23 FLOP | 48B | Open weights (unrestricted) |
| DeepSeekMath 7B | DeepSeek, Tsinghua University, Peking University | Feb 5, 2024 | 10^23 FLOP | 7B | Open weights (restricted use) |
| Qwen-7B | Alibaba | Sep 28, 2023 | 10^23 FLOP | 7B | Open weights (restricted use) |
| Evo 2 7B | Arc Institute, Stanford University, NVIDIA, Liquid, University of California (UC) Berkeley, Goodfire, Columbia University, University of California San Francisco | Feb 19, 2025 | 10^23 FLOP | 7B | Open weights (unrestricted) |
| Luminous-extended | Aleph Alpha | Aug 15, 2022 | 10^23 FLOP | 30B | API access |
| Phi-4 Mini | Microsoft | Mar 3, 2025 | 10 x 10^22 FLOP | 3.8B | Open weights (unrestricted) |
| GigaChat Lite (GigaChat-20B-A3B) | Sber | Dec 13, 2024 | 9.9 x 10^22 FLOP | 20B | Open weights (unrestricted) |
| MP4 | 310.ai | Jul 15, 2024 | 9.7 x 10^22 FLOP | — | Unreleased |
| Gemma 3 4B | Google DeepMind | Mar 12, 2025 | 9.6 x 10^22 FLOP | 4B | Open weights (restricted use) |
| Palmyra Large 20B | Writer | Mar 1, 2023 | 9.6 x 10^22 FLOP | 20B | Open weights (unrestricted) |
| EXAONE 3.5-R 2.4B | LG AI Research | Mar 14, 2025 | 9.5 x 10^22 FLOP | 2.4B | Unreleased |
| EXAONE Deep 2.4B | LG AI Research | Mar 16, 2025 | 9.4 x 10^22 FLOP | 2.4B | Open weights (non-commercial) |
| EXAONE 3.5 2.4B | LG AI Research | Dec 9, 2024 | 9.4 x 10^22 FLOP | 2.4B | Open weights (non-commercial) |
| Baichuan 1-13B | Baichuan | Jul 11, 2023 | 9.4 x 10^22 FLOP | 13.3B | Open weights (restricted use) |
| GPT-NeoX-20B | EleutherAI | Feb 9, 2022 | 9.3 x 10^22 FLOP | 20B | Open weights (unrestricted) |
| Trillion-7B | Trillion Labs | Apr 21, 2025 | 9.3 x 10^22 FLOP | 7B | Open weights (unrestricted) |
| ALLaM 7B | Saudi Data and Artificial Intelligence Authority | May 21, 2024 | 9 x 10^22 FLOP | 7B | Open weights (unrestricted) |
| Zamba2-7B | Zyphra | May 26, 2024 | 8.8 x 10^22 FLOP | 7B | Open weights (unrestricted) |
| phi-3.5-Vision | Microsoft | Apr 23, 2024 | 8.8 x 10^22 FLOP | 4.2B | Open weights (unrestricted) |
| EXAONE 4.0 (1.2B) | LG AI Research | Jul 15, 2025 | 8.7 x 10^22 FLOP | 1.2B | Open weights (non-commercial) |
| TeleChat-12B | China Telecom | Apr 1, 2024 | 8.6 x 10^22 FLOP | 12B | Open weights (restricted use) |
| SambaLingo-Thai-Chat (7B) | SambaNova Systems, Inc | Apr 8, 2024 | 8.6 x 10^22 FLOP | 7B | Open weights (unrestricted) |
| StarCoder | Hugging Face, ServiceNow, Northeastern University, Mila - Quebec AI (originally Montreal Institute for Learning Algorithms), Carnegie Mellon University (CMU), Johns Hopkins University, Leipzig University, ScaDS.AI, Queen Mary University of London, Roblox, Sea AI Lab, Technion - Israel Institute of Technology, Monash University, CSIRO, Data61, McGill University, Saama, University of British Columbia (UBC), Massachusetts Institute of Technology (MIT), Technical University of Munich, IBM, University of Vermont, UnfoldML, SAP, University of Notre Dame, Columbia University, New York University (NYU), University of Allahabad, Discover Dollar, Toloka, Telefonica, Stanford University, Weizmann Institute of Science, Alan Turing Institute, Wellesley College, EleutherAI, Forschungszentrum Julich | May 9, 2023 | 8.5 x 10^22 FLOP | 15.5B | Open weights (restricted use) |
