list all the ai labs started in 5 years, grouped by year
A web-wide census of 90 independent organizations founded in the calendar years 2022–2026 (through 2 September 2026) whose central work is core AI research or general-purpose, foundation, generative, reasoning, world, scientific, or embodied AI models. One square per lab below; every row links its cited source.
Each square is one lab, stacked by founding year; square size grows with corroboration.corroborated (2+ supporting pages in this scan)single-source / verify. Source count is corroboration found in this scan — not total web mentions, and not a quality score.
| Lab | Focus | Founders | HQ | Source |
|---|---|---|---|---|
| Adept AI Labscorroborated · 11 | ACT-1 (Action Transformer)—foundation model trained to interact with computers, browsers, and enterprise software through natural language commands. | David Luan, Ashish Vaswani, Niki Parmar | San Francisco, California, USA | eboona.com |
| Inflection AIcorroborated · 7 | develop AI software products that make it easier for humans to communicate with computers | Reid Hoffman and Mustafa Suleyman | Silicon Valley, California, U.S. | reuters.com |
| ElevenLabs Inc.corroborated · 5 | Voice AI, text-to-speech, voice agents | Piotr Dąbkowski and Mateusz Staniszewski | London, UK | northflank.com |
| Perplexity AI, Inc.corroborated · 5 | Artificial intelligence, search engine | Aravind Srinivas, Denis Yarats, Johnny Ho, Andy Konwinski | San Francisco | clarigital.com |
| Reka AIcorroborated · 3 | Dani Yogatama, Yi Tay, Qi Liu, and Aaron Ng | Singapore | nextomoro.com | |
| Together AIcorroborated · 3 | Vipul Ved Prakash, Ce Zhang, Percy Liang, Chris Re | San Francisco, CA | xyzeo.com | |
| EquiLibre Technologiescorroborated · 2 | reinforcement learning applied to trading stocks and financial markets | Martin Schmid, Rudolf Kadlec, and Matej Moravcik | Prague, Czechia | techcrunch.com |
| Essential AIcorroborated · 2 | human-machine partnership | Ashish Vaswani | profile.lessie.ai | |
| Coheresingle-source / verify | Enterprise LLMs | theplanettools.ai | ||
| Imbuesingle-source / verify | mastersofscale.com | |||
| Noetiksingle-source / verify | AI to tackle one of the toughest challenges in cancer treatment: finding the right targets and understanding how drugs will work on different patients | Jacob Rinaldi and Ron Alfa | San Francisco Bay Area | biopharmatrend.com |
| Lab | Focus | Founders | HQ | Source |
|---|---|---|---|---|
| Sakana AIcorroborated · 25 | AI research lab focused on model research and efficiency. | David Ha, Llion Jones | Tokyo, Japan | presenc.ai |
| Mistral AIcorroborated · 20 | focuses on developing language models with varying capabilities and strengths | Arthur Mensch and ex-DeepMind | Paris, France | presenc.ai |
| Moonshot AIcorroborated · 17 | large language models | Yang Zhilin, Zhou Xinyu, and Wu Yuxin | China | en.wikipedia.org |
| SpaceXAI LLC (formerly xAI)corroborated · 16 | generative AI | Elon Musk | Stanford Research Park, Palo Alto, California, United States | en.wikipedia.org |
| DeepSeek AIcorroborated · 10 | Open-source foundation model developer. | Liang Wenfeng | China | nextomoro.com |
| Poolside AIcorroborated · 9 | Automated code creation, testing, and project scaffolding | Jason Warner, Eiso Kant, Pengming Wang, Nikolay Zinov | United States (with a LinkedIn location in San Francisco, California) | opensourcemischief.blog |
| 01.AIcorroborated · 7 | developing smaller, industry-specific AI models while balancing both open-source and proprietary solutions | Kai-Fu Lee | Beijing, China | topbots.com |
| Baichuan Intelligencecorroborated · 7 | AI-driven solutions in healthcare, education, and finance | Wang Xiaochuan | Beijing, China | topbots.com |
| EvolutionaryScalecorroborated · 4 | develops AI models to generate novel proteins for scientific research | Alexander Rives, Tom Secru, and Sal Candido | New York, NY, United States | doi.bio |
| Latent Labscorroborated · 3 | generative artificial intelligence (AI) models to accelerate drug discovery and research | Dr Simon Kohl | London and San Francisco | siliconrepublic.com |
| StepFuncorroborated · 3 | AIGC large models | Jiang Daxin | Shanghai, China | wicinternet.org |
| Skild AIcorroborated · 2 | building the Skild Brain, the world's first omni-bodied robotics foundation model | Deepak Pathak and ex-CMU robotics | Pittsburgh, PA, USA | presenc.ai |
| AAIsingle-source / verify | Artificial Expert Intelligence (AEI) systems intended to perform at the level of leading human experts in science, mathematics, and adjacent technical fields | Amnon Shashua and Shai Shalev-Shwartz | Ramat Gan, Israel | nextomoro.com |
| Agibotsingle-source / verify | humanoid-robotics | Peng Zhihui | Shanghai, China | nextomoro.com |
| Amazon AGIsingle-source / verify | nextomoro.com | |||
| Basaltsingle-source / verify | frontier AI systems, open model weights and safety research | Chen Mingzhi and Li Wenjuan | Beijing, China | basaltlabs.org |
| Cognition AIsingle-source / verify | Autonomous AI software engineers | San Francisco, California, USA | wellows.com | |
| EngineAIsingle-source / verify | gait-and-motion-control engineering on the SE01 and PM01 platforms | Shenzhen, China | nextomoro.com | |
| Factorysingle-source / verify | San Francisco, United States | tracxn.com | ||
| Harmonicsingle-source / verify | Vlad Tenev, Tudor Achim | reuters.com | ||
| Helicalsingle-source / verify | bio foundation models for DNA and RNA data | Rick Schneider, Mathieu Klop, and Maxime Allard | Luxembourg | biopharmatrend.com |
| Lingyiwanwusingle-source / verify | hurun.net | |||
| Odysseysingle-source / verify | Oliver Cameron and Jeff Hawke | techcrunch.com | ||
| Sierrasingle-source / verify | Conversational AI agents for enterprise customer service | Bret Taylor and Clay Bavor | San Francisco, California, USA | wellows.com |
| Lab | Focus | Founders | HQ | Source |
|---|---|---|---|---|
| Black Forest Labscorroborated · 30 | image generation model family | Robin Rombach, Andreas Blattmann, Patrick Esser, and Dominik Lorenz | Freiburg, Germany | siliconvalleyinvestclub.com |
| Reflection AIcorroborated · 27 | autonomous coding agents and the development of an open-weights frontier language model | Misha Laskin and Ioannis Antonoglou | San Francisco, United States | nextomoro.com |
| World Labscorroborated · 25 | Spatial intelligence and 3D world models | Fei-Fei Li and team | San Francisco, California, USA | presenc.ai |
| Safe Superintelligence Inc.corroborated · 15 | Developing safe superintelligence | Ilya Sutskever, Daniel Gross, Daniel Levy | Palo Alto, California, USA & Tel Aviv, Israel | presenc.ai |
| Physical Intelligence (Pi)corroborated · 11 | foundational software for robots | Karol Hausman, Sergey Levine, and others | San Francisco, California | presenc.ai |
| Genesis AIcorroborated · 10 | to build a general-purpose model that will enable robots to automate a wide range of repetitive tasks, from lab work to housekeeping | Zhou Xian and Théophile Gervet | Silicon Valley and Paris | techcrunch.com |
| Goodfire AIcorroborated · 7 | AI interpretability research | Eric Ho, Daniel Balsam, and Tom McGrath | San Francisco, United States | nextomoro.com |
| Scout AIcorroborated · 3 | frontier AI for defense, specifically Vision-Language-Action foundation models for defense robotics | Colby Adcock and Collin Otis | Sunnyvale, CA | yespress.io |
| Dyna Roboticscorroborated · 2 | Develops AI-powered solutions for business automation and productivity in robotics contexts. | Redwood City | siliconvalleyinvestclub.com | |
| Fundamental Research Labscorroborated · 2 | AI Lab offering foundational models | techcrunch.com | ||
| General Intuitioncorroborated · 2 | Pim de Witte, Adam Jelley, Eloi Alonso, and Vincent Micheli | New York City, United States | nextomoro.com | |
| H Companycorroborated · 2 | agentic AI products including Runner H, Surfer H, and Tester H, along with the open-weights Holo-1 visual-language model | ex-DeepMind agent team | Paris, France | presenc.ai |
| Rhodacorroborated · 2 | Develops robotic systems using video pre-training for industrial automation. | Palo Alto | siliconvalleyinvestclub.com | |
| Avirosingle-source / verify | We focus on tasks involving thousands of tool steps for coding, computer use, and knowledge work. | San Francisco | ycombinator.com | |
| Converge Biosingle-source / verify | generative AI trained on molecular data to help pharma and biotech companies develop drugs faster | Dov Gertz | Boston and Tel Aviv | techcrunch.com |
| Deep Cogitosingle-source / verify | to develop AI models that possess the ability to autonomously improve themselves | Drishan Arora and Dhruv Malrana | siliconangle.com | |
| Generalistsingle-source / verify | techfundingnews.com | |||
| Inceptionsingle-source / verify | large language models | a Stanford University team | reuters.com | |
| NewMes Inc.single-source / verify | AI-powered hyper-magnification MRI platform, AI-driven ultra-magnification, zero-shot self-learning AI, and medical AI algorithms | Seoul, South Korea | newmes.io | |
| Perceptronsingle-source / verify | techcrunch.com | |||
| RLWRLDsingle-source / verify | Platform focused on building foundational AI models | Seoul, South Korea | tracxn.com | |
| Sesamesingle-source / verify | conversational speech models and smart-glasses hardware | Brendan Iribe and Ankit Kumar | nextomoro.com | |
| Standard Intelligencesingle-source / verify | nextomoro.com | |||
| Tilde Researchsingle-source / verify | mechanistic interpretability, new architectures, and pretraining science | Dhruv Pai | Palo Alto, United States | techinterview.org |
| Transformer Labsingle-source / verify | Ali Asaria and Tony Salomone | Kitchener-Waterloo, Canada | betakit.com |
| Lab | Focus | Founders | HQ | Source |
|---|---|---|---|---|
| Thinking Machines Labscorroborated · 40 | building a model that processes your input and generates a response at the same time, so it’s more like a phone call than a text chain | Mira Murati | San Francisco, CA | techcrunch.com |
| Ineffable Intelligencecorroborated · 33 | building a "superlearner," an AI system designed to acquire knowledge and skills through its own experience and interaction with environments rather than through large volumes of human-generated data | David Silver | Britain | prismnews.com |
| Advanced Machine Intelligence Labscorroborated · 23 | World models, specifically designed to reason about the physical world. | Yann LeCun and Alexandre LeBrun | Paris, France | nextomoro.com |
| Periodic Labscorroborated · 7 | autonomous labs | Liam Fedus and Ekin Dogus Cubuk | San Francisco, United States | cnb.cx |
| Ricursive Intelligencecorroborated · 7 | chip design | Anna Goldie and Azalia Mirhoseini | Palo Alto, California | cnb.cx |
| Prior Labscorroborated · 6 | Tabular Foundation Models (TFMs) for structured business data | Frank Hutter, Noah Hollmann, Sauraj Gambhir | Freiburg, Germany | ukpostobserver.com |
| Humans&corroborated · 4 | reinforcement learning | San Francisco, United States | cnb.cx | |
| Adaption Labscorroborated · 2 | adaptive data, adaptive intelligence, and adaptive interfaces | Sara Hooker and Sudip Roy | San Francisco, United States | nextomoro.com |
| Flapping Airplanescorroborated · 2 | AI research lab | techcrunch.com | ||
| ARYA Labssingle-source / verify | Dr. Seth Dobrin and Łukasz Chmiel | New York, United States | einnews.com | |
| Blur Labssingle-source / verify | open-source AI foundation models that can reconstruct higher-quality images from limited or noisy data | Andrew Wang, Julián Tachella, Matthieu Terris | Paris, France | engineering-designer.com |
| Cormasingle-source / verify | defensive cybersecurity | Alon Pluda | Tel Aviv and San Francisco | theaiinsider.tech |
| Eloriansingle-source / verify | nextomoro.com | |||
| Good Start Labssingle-source / verify | teach AI models to play games and generate high-quality reinforcement learning data for frontier labs | Alex Duffy and Tyler Marques | every.to | |
| Gradiumsingle-source / verify | audio language AI models designed to deliver voice at scale with ultra-low latency | Neil Zeghidour | Paris, France | techcrunch.com |
| Harksingle-source / verify | building models and hardware for an AI personal assistant | Brett Adcock | techcrunch.com | |
| Idlersingle-source / verify | specializes in developing reinforcement learning environments for training AI models based on real-world coding scenarios | San Francisco, California | builtin.com | |
| Integrous Researchsingle-source / verify | safety-focused AI research | Arvid Lunnemark | en.wikipedia.org | |
| Labs Companies, Inc.single-source / verify | Frontier Intelligences, Human-enhancement technologies | Durand F. Davis Jr., Maya E. Davis | Silicon Beach - Santa Monica, CA | labscompanies.com |
| Meta Superintelligence Labssingle-source / verify | AI division | Alexandr Wang | en.wikipedia.org | |
| Ndeasingle-source / verify | artificial general intelligence through deep learning-guided program synthesis | François Chollet and Mike Knoop | San Francisco, United States | nextomoro.com |
| UMAsingle-source / verify | general-purpose mobile and humanoid robots capable of human-level dexterity and understanding of the physical world | Paris | linkedin.com |
| Lab | Focus | Founders | HQ | Source |
|---|---|---|---|---|
| Recursive Superintelligencecorroborated · 17 | AI systems that can recursively improve themselves | Tim Rocktäschel | United Kingdom | cnb.cx |
| Prentiscorroborated · 10 | computer use models | Ritankar Das, Reid Hoffman, and Marc Pincus | genainews.tech | |
| Core Automationcorroborated · 2 | Frontier AI lab building continual learning systems | Jerry Tworek, Julia Villagra | San Francisco, CA | uuaa.net |
| Mirendilcorroborated · 2 | AI for biology and materials science | Behnam Neyshabur and Harsh Mehta | San Francisco, United States | nextomoro.com |
| Inherentsingle-source / verify | Tantum Collins, Edward Hughes, Louis Kirsch, Kaloyan Aleksiev | London, United Kingdom | appliedmethods.ai | |
| Oak Labsingle-source / verify | continuously learning AI agents using the OaK (Options and Knowledge) architecture | Richard Sutton and Khurram Javed | Canada | mlq.ai |
| Prentisssingle-source / verify | computational models | Retankar Das, Reid Hoffman, and Mark Pincus | yourai.pro | |
| Walden Roboticssingle-source / verify | pioneers in robotics and AI from Toyota Research Institute, MIT, Stanford, and Amazon | Cambridge, Mass. | innovationopenlab.com |
Data: web-wide scan canonicalized to 90 candidate AI labs founded 2022–2026, current through 2026-09-02. One row per lab with founding year, focus, founders, headquarters, count of distinct supporting pages found, and one cited source. Blank cells are unresolved details. Edge cases may include lab-like AI companies, robotics/model companies, scientific-AI labs, and corporate research units; low-corroboration rows can be false positives or date ambiguities. Nothing was cut for space.