 Alibaba QwenCompany lab · Hangzhou, ChinaHangzhou, ChinaThe Qwen family, the most downloaded open-weight models, from Alibaba Cloud's Tongyi lab. | The Qwen family, the most downloaded open-weight models, from Alibaba Cloud's Tongyi lab. | 2023 | Qwen3.8 Max (0902) | 1 |
 Amazon AGI and Amazon ScienceCompany lab · Seattle, United StatesSeattle, United StatesThe Amazon Nova models and Nova Act agents, plus Chronos forecasting models from Amazon Science. | The Amazon Nova models and Nova Act agents, plus Chronos forecasting models from Amazon Science. | n/a | Amazon Nova | 0 |
 Apple Machine Learning ResearchCompany lab · Cupertino, United StatesCupertino, United StatesOn-device models for Apple Intelligence, the MLX framework for Apple silicon, and studies of reasoning limits. | On-device models for Apple Intelligence, the MLX framework for Apple silicon, and studies of reasoning limits. | n/a | None | 0 |
 Baidu (ERNIE)Company lab · Beijing, ChinaBeijing, ChinaThe ERNIE models and PaddlePaddle, and earlier Deep Speech 2 from its Silicon Valley lab. | The ERNIE models and PaddlePaddle, and earlier Deep Speech 2 from its Silicon Valley lab. | n/a | Ernie 5.1 | 0 |
 ByteDance SeedCompany lab · ChinaChinaByteDance's model team behind Doubao, the UI-TARS computer-use agents and Seedance video generation. | ByteDance's model team behind Doubao, the UI-TARS computer-use agents and Seedance video generation. | n/a | None | 0 |
 Databricks Mosaic ResearchCompany lab · San Francisco, United StatesSan Francisco, United StatesOpen models (MPT, DBRX) and training at scale, from the MosaicML team Databricks bought in 2023; Spark lineage. | Open models (MPT, DBRX) and training at scale, from the MosaicML team Databricks bought in 2023; Spark lineage. | 2023 | None | 1 |
 Google DeepMindCompany lab · London, United KingdomLondon, United KingdomAlphaGo and AlphaFold, deep reinforcement learning, and Google's Gemini models. | AlphaGo and AlphaFold, deep reinforcement learning, and Google's Gemini models. | 2010 | Gemini 3.8 Flash | 3 |
 Google ResearchCompany lab · Mountain View, United StatesMountain View, United StatesThe Transformer, BERT and T5, plus MapReduce and TensorFlow: much of the base of modern AI. | The Transformer, BERT and T5, plus MapReduce and TensorFlow: much of the base of modern AI. | 2000 | None | 4 |
 IBM ResearchCompany lab · Yorktown Heights, United StatesYorktown Heights, United StatesDeep Blue and Watson, relational databases and quantum computing, and today the open Granite models. | Deep Blue and Watson, relational databases and quantum computing, and today the open Granite models. | 1945 | IBM Granite | 0 |
 Isomorphic LabsCompany lab · London, United KingdomLondon, United KingdomAI-first drug discovery spun out of DeepMind; co-developed AlphaFold 3. | AI-first drug discovery spun out of DeepMind; co-developed AlphaFold 3. | 2021 | None | 0 |
 Meta FAIRCompany lab · New York, United StatesNew York, United StatesPyTorch, the open Llama models, Segment Anything and self-supervised vision (DINO). | PyTorch, the open Llama models, Segment Anything and self-supervised vision (DINO). | 2013 | None | 4 |
 Meta Superintelligence LabsCompany lab · Menlo Park, United StatesMenlo Park, United StatesMeta's frontier model effort, formed in 2025 under Alexandr Wang; it builds the Muse models and now houses FAIR. | Meta's frontier model effort, formed in 2025 under Alexandr Wang; it builds the Muse models and now houses FAIR. | 2025 | Muse Spark 1.3 | 0 |
 Microsoft AICompany lab · Redmond, United StatesRedmond, United StatesCopilot for consumers and Microsoft's own MAI models, led by Mustafa Suleyman since 2024. | Copilot for consumers and Microsoft's own MAI models, led by Mustafa Suleyman since 2024. | 2024 | None | 0 |
 Microsoft ResearchCompany lab · Redmond, United StatesRedmond, United StatesResNet, LoRA, DeepSpeed and the small Phi models, from a lab spanning systems, theory and AI. | ResNet, LoRA, DeepSpeed and the small Phi models, from a lab spanning systems, theory and AI. | 1991 | Phi-4 | 4 |
 NVIDIA ResearchCompany lab · Santa Clara, United StatesSanta Clara, United StatesStyleGAN, Megatron-LM for training giant models, Instant NeRF, and the open Nemotron models. | StyleGAN, Megatron-LM for training giant models, Instant NeRF, and the open Nemotron models. | 2006 | Nemotron-4 340B | 3 |
 Salesforce AI ResearchCompany lab · United StatesUnited StatesBLIP vision-language models, CodeT5, and xLAM action models for agents. | BLIP vision-language models, CodeT5, and xLAM action models for agents. | 2016 | None | 0 |
 Snowflake AI ResearchCompany lab · Menlo Park, United StatesMenlo Park, United StatesArctic: open embedding models, the Arctic LLM, SwiftKV inference and text-to-SQL research. | Arctic: open embedding models, the Arctic LLM, SwiftKV inference and text-to-SQL research. | n/a | None | 3 |
 Tencent Hy (Hunyuan)Company lab · Shenzhen, ChinaShenzhen, ChinaTencent's Hunyuan models, with open releases in MoE language, video (HunyuanVideo) and 3D generation. | Tencent's Hunyuan models, with open releases in MoE language, video (HunyuanVideo) and 3D generation. | 2023 | None | 0 |
 xAI (SpaceXAI)Company lab · San Francisco Bay Area, United StatesSan Francisco Bay Area, United StatesThe Grok models and the Colossus supercomputer in Memphis; now a division of SpaceX. | The Grok models and the Colossus supercomputer in Memphis; now a division of SpaceX. | 2023 | Grok 4.7 | 0 |
