Large Language Models Companies

Apple AI

Cupertino, United States

Apple AI focuses on the research and development of on-device artificial intelligence, primarily through its MLX framework – an Apple silicon-optimized toolkit for developing and deploying large language models. Targeting AI researchers and developers, MLX enables private, efficient LLM experimentation and deployment directly on Apple hardware, bypassing reliance on cloud infrastructure. This positions Apple AI as a provider of both foundational ML technology and a secure, localized AI development environment.

Microsoft AI

Redmond, United States

Microsoft AI provides a comprehensive suite of artificial intelligence services and products, including the Copilot assistant and Azure AI platform. Leveraging large language models developed in partnership with OpenAI, they offer generative AI and cloud-based AI solutions for enterprise applications. Their focus is on enabling businesses to integrate AI capabilities for increased productivity, innovation, and sustainability initiatives.

Google Cloud AI

Mountain View, United States

Google Cloud AI provides a comprehensive suite of cloud-based artificial intelligence services for enterprise customers. Their core offering, Vertex AI, is a unified platform enabling the full lifecycle of machine learning development – from model building and tuning to deployment and MLOps. Google Cloud AI differentiates itself through decades of internal AI research – including the development of transformer models – and a focus on enabling customizable, production-ready AI agents for business automation.

Google AI

Mountain View, United States

and general knowledge: Google AI develops cutting-edge large language models like Gemini, alongside the widely-adopted TensorFlow machine learning framework and supporting infrastructure. Their key innovations center on multimodal AI – enabling models to process and understand text, images, audio, and video – as demonstrated by Gemini’s advanced reasoning and creative capabilities. Targeting developers, researchers, and general consumers, Google AI integrates these technologies across numerous Google products and recently launched the Gemini API for broader access and application development.

Amazon AI

Seattle, United States

Amazon AI provides a comprehensive suite of cloud-based artificial intelligence services and infrastructure through AWS, including the development of foundation models like Titan and the SageMaker AI platform. Their core offering centers on enabling organizations to build, train, and deploy agentic AI applications at scale, leveraging specialized infrastructure and a robust data foundation. Targeting businesses across all industries, Amazon AI differentiates itself by offering end-to-end capabilities – from model development to data governance – with a focus on cost optimization and enterprise-grade security.

Salesforce AI

San Francisco, United States

Salesforce AI delivers embedded artificial intelligence capabilities across the Salesforce Customer 360 platform, most notably through its Einstein GPT and Einstein Copilot offerings. Leveraging large language models and proprietary machine learning, Salesforce AI provides generative AI features like automated email generation, call summarization (via Sales AI), personalized marketing journeys, and AI-powered customer service tools like Agentforce – all integrated with CRM data. These solutions are targeted towards enterprise businesses seeking to enhance sales, service, marketing, and commerce operations, and have enabled features like automated knowledge base creation and the ability to integrate custom LLMs via the Einstein Trust Layer.

IBM watsonx

Armonk, United States

IBM watsonx is an enterprise AI platform offering foundation models – including open-source options – and governance tools designed to accelerate the deployment of generative AI. The platform focuses on enabling businesses to build, deploy, and monitor AI assistants and agents across hybrid cloud environments. watsonx targets large organizations seeking to integrate AI into core workflows while maintaining control, compliance, and data security.

Snowflake

Bozeman, United States

Snowflake provides a cloud-based data platform enabling organizations to store, process, and analyze data at scale. Their core offering, the AI Data Cloud, incorporates features like Cortex AI to facilitate the development and deployment of custom large language models (LLMs) directly within the platform. Snowflake targets data-intensive enterprises seeking to break down data silos and accelerate AI innovation without data movement or complex infrastructure management.

Baidu AI

Beijing, China

Baidu is China's leading AI company with ERNIE LLM, Apollo autonomous driving, and PaddlePaddle deep learning framework.

OpenAI

San Francisco, United States

OpenAI is an AI research and deployment company dedicated to ensuring that artificial general intelligence benefits all of humanity. Creator of GPT-4, ChatGPT, DALL-E, and Sora.

Anthropic

San Francisco, United States

Anthropic is an AI safety company working to build reliable, interpretable, and steerable AI systems. Creator of Claude, focused on Constitutional AI and harmlessness research.

Duolingo

Pittsburgh, United States

Duolingo is a US-based EdTech company offering free, gamified language learning across a wide range of languages. Their core offering is enhanced by AI-powered personalization, most notably through Duolingo Max which integrates OpenAI’s GPT-4 to provide users with realistic conversation practice and detailed feedback. Targeting a broad consumer base of language learners, Duolingo differentiates itself through accessibility, a science-backed methodology, and increasingly sophisticated AI-driven features.

xAI

San Francisco, United States

xAI is a US-based artificial intelligence company focused on developing and deploying large language models with a stated mission of advancing scientific discovery. Their primary product is Grok, a generative AI model accessible via API, web, and mobile platforms, offering enhanced speed, precision, and multilingual capabilities. xAI targets both individual users seeking an AI assistant and developers looking to integrate advanced LLM functionality into their applications, as demonstrated by their API offerings and recent partnership with the government of El Salvador.

Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, builds more understandable and customizable AI systems. Raised a record $2B seed round.

Reflection AI

San Francisco, United States

Reflection AI is developing large language models focused on reasoning and long-context understanding, with a core product currently in limited access called “Reflect.” The company differentiates itself through a novel architecture designed to improve model reliability and reduce “hallucinations,” aiming for more trustworthy generative AI outputs. Backed by a $2 billion Series B led by Nvidia valuing the company at $8 billion, Reflection AI is positioning itself as a key player in the open intelligence movement, making its models accessible for research and development.

Inflection AI

Palo Alto, United States

Inflection AI builds personal AI assistants designed to be helpful, harmless, and emotionally intelligent. Creator of Pi.

SambaNova Systems

Palo Alto, United States

SambaNova Systems develops a full-stack AI platform, including DataScale processors (RDUs) and the Samba-1 model suite, designed to accelerate AI inference and fine-tuning. The company offers both cloud-based (SambaCloud) and on-premise (SambaStack) deployment options, targeting enterprises and governments with demanding data security and performance requirements. SambaNova positions itself as a high-performance, energy-efficient alternative to GPU-based AI infrastructure, particularly for large language models and sovereign AI initiatives.

Moonshot AI

Beijing, China

Moonshot AI is a Chinese artificial intelligence company specializing in the development of large language models. Their core product, Kimi, is a long-context LLM distinguished by its capacity to process exceptionally large input sequences – reportedly millions of tokens – enabling advanced applications in complex document analysis and information retrieval. This capability positions Moonshot AI to serve organizations requiring processing of extensive datasets, such as legal firms, research institutions, and financial analysts.

Safe Superintelligence (SSI), founded by OpenAI co-founder Ilya Sutskever, focuses exclusively on building safe superintelligent AI.

Crusoe Energy

Denver, United States

Crusoe Energy provides scalable AI infrastructure and cloud compute services, specializing in deployments for large-context AI models. Their core offering, Crusoe Cloud, leverages proprietary MemoryAlloy technology and optimized hardware – including the latest NVIDIA & AMD GPUs – to deliver accelerated inference speeds and reduced costs. Crusoe targets organizations requiring high-performance, reliable AI compute, uniquely powered by stranded natural gas and renewable energy sources to offer a sustainable infrastructure solution.

Cerebras Systems

Sunnyvale, United States

Cerebras Systems develops AI hardware, specifically the Wafer Scale Engine, a large-scale chip designed to accelerate deep learning workloads. Unlike traditional GPUs, Cerebras’ technology aims to significantly reduce the time and cost associated with training complex AI models. Their target market is organizations requiring high-performance computing for demanding AI applications, such as large language models and scientific computing.

Minimax

Shanghai, China

aiming for the requested detail and professionalism: Minimax focuses on building generative AI models for content creation, demonstrated by its Talkie AI conversational avatar platform and Hailuo, a text-to-video generation tool capable of producing 1080p videos. The company differentiates itself through innovations in AI-driven digital human creation and realistic video synthesis, leveraging large language models and diffusion models for high-fidelity output. Minimax has gained traction in the Chinese market with applications targeting entertainment, education, and virtual influencers, and recently secured significant funding to expand its multimodal AI capabilities.

Minimax

Shanghai, China

Minimax builds multimodal AI including text, voice, and video generation models for consumer and enterprise applications.

Mistral AI is a French AI company developing efficient open-weight large language models. Known for Mistral 7B, Mixtral, and enterprise solutions.

Aleph Alpha

Heidelberg, Germany

Aleph Alpha builds sovereign, trustworthy AI for European enterprises and governments. Known for Luminous models and focus on explainability.

Chegg

Santa Clara, United States

Chegg is a U.S.-based EdTech company providing on-demand academic support to students. Their core offering is CheggMate, an AI-powered assistant that delivers step-by-step solutions and expert answers to homework questions, alongside textbook rentals and writing tools. Chegg primarily serves high school and college students seeking accessible, 24/7 assistance with coursework and exam preparation.

Cohere

Toronto, Canada

Cohere provides enterprise AI solutions with industry-leading LLMs optimized for business applications, including Command and Embed models.

Retool

San Francisco, United States

Retool is a developer platform enabling businesses to rapidly build and deploy custom internal tools. Their core offering is a unified engine that connects to diverse data sources – including databases, APIs, and Large Language Models – with a focus on production-grade reliability. Retool targets organizations seeking to overcome engineering bottlenecks and efficiently create purpose-built software for operational workflows, data management, and AI integration without extensive coding.

NotCo

Santiago, Chile

NotCo is a food technology company that leverages a proprietary AI platform, Giuseppe, to accelerate and optimize the development of plant-based food products. Their AI utilizes a high-fidelity data corpus and multi-objective optimization to rapidly explore formulation possibilities, addressing constraints related to cost, nutrition, sensory appeal, and manufacturability. NotCo targets food manufacturers seeking to innovate and scale plant-based alternatives more efficiently than traditional R&D methods allow.

Zhipu AI

Beijing, China

Zhipu AI is a Chinese artificial intelligence company specializing in the development of large language models (LLMs), most notably the GLM-130B and ChatGLM series. They provide both open-source and API access to their models, including those optimized for code generation (CodeGeeX) and visual understanding (CogView). Zhipu AI targets enterprise clients and developers in China, offering solutions for applications like chatbots, content creation, and AI-assisted workflows.

VAST Data

New York, United States

VAST Data provides an AI Operating System that unifies storage, database, and compute resources into a single, orchestrated platform. Built on their Disaggregated, Autonomous, Storage Engine (DASE) architecture, the system is designed to eliminate data bottlenecks and deliver terabyte-per-second performance to large-scale GPU clusters. VAST Data targets enterprises deploying demanding, data-intensive AI and agentic computing workloads, offering a solution focused on scalability, performance, and reduced total cost of ownership.

Physical Intelligence is building foundation models for physical AI, enabling robots to learn general-purpose manipulation skills.

Weka

Campbell, United States

Weka provides a high-performance data platform, NeuralMesh, designed to accelerate AI and machine learning workloads. Their technology utilizes a distributed, parallel file system to extend GPU memory by up to 1000x and significantly reduce AI response times – demonstrated by a 20x reduction in time-to-first-token. Weka targets organizations building and scaling demanding agentic AI applications, offering a solution to maximize GPU utilization and improve the economics of AI infrastructure, with deployment options including Oracle Cloud.

Fractal AI

Mumbai, India

Fractal Analytics is a global provider of artificial intelligence and analytics solutions focused on enterprise applications. The company specializes in end-to-end AI implementation, and has recently been selected by the Indian government to develop the nation’s first Large Reasoning Model (LRM). Fractal targets large global enterprises seeking to integrate AI for improved decision-making and operational efficiency.

Redis

Mountain View, United States

Redis provides a fully-managed, in-memory data platform optimized for real-time applications, including those leveraging artificial intelligence. Their core offering is Redis Stack, which incorporates vector databases and semantic search capabilities to accelerate AI workloads like chatbots and LLM-powered agents. Redis targets developers seeking to reduce latency and infrastructure costs associated with AI applications by providing a high-performance caching and data storage layer.

Gupshup

San Francisco, United States

Gupshup is a conversational AI company providing a platform for businesses to deploy autonomous AI Agents across messaging channels. Their core technology centers on a proprietary, fine-tuned Large Language Model (LLM) called ACE, enabling personalized and scalable customer interactions. Gupshup targets enterprises seeking to automate sales, marketing, and customer support functions through AI-driven conversations, while also offering AI tools to enhance human agent productivity.

AI21 Labs

Tel Aviv, Israel

AI21 Labs develops advanced large language models, including the Jurassic-2 and Jamba series, designed for enterprise applications. Their key innovation lies in long-context language modeling – Jamba specifically offers 30% reduced compute for faster inference on extended texts – alongside flexible deployment options including on-premise and hybrid cloud setups to address data privacy and compliance needs. AI21 Labs serves a range of industries, including fintech and academic research – demonstrated through partnerships leveraging their Retrieval-Augmented Generation (RAG) capabilities for cited, scholarly answers and around-the-clock service availability.

Sky Mavis

Ho Chi Minh City, Vietnam

Sky Mavis is a Vietnam-based game development studio and blockchain technology company building play-to-earn games and supporting infrastructure. Their flagship product, Axie Infinity, utilizes non-fungible tokens (NFTs) and blockchain technology to create a player-owned digital ecosystem. Sky Mavis also develops Ronin, an Ethereum Virtual Machine (EVM) blockchain specifically designed to scale games with player-owned economies and reduce transaction fees.

StepFun develops multimodal AI models specializing in vision-language understanding for a broad range of personal productivity tasks. Their core product is an AI assistant capable of knowledge retrieval, language learning, content creation (text & code), and general information processing. Targeting individual users in China, StepFun aims to improve efficiency across work, study, and daily life through accessible AI tools.

Skild AI

Pittsburgh, United States

Skild AI develops scalable robot brain models that work across different robots and tasks. Raised $300M Series A at $1.5B valuation.

Sigma Computing

San Francisco, United States

Sigma Computing provides a cloud-native business intelligence and analytics platform that directly integrates with cloud data warehouses like Databricks. Their core technology leverages large language models (LLMs) to enable AI-powered data exploration, automated insights (“Ask Sigma”), and the creation of live, data-driven applications within a familiar spreadsheet interface. Sigma targets enterprises seeking to unify data access, accelerate analytics workflows, and empower users with self-service BI without data silos or stale exports.

K Health

New York, United States

K Health is a U.S.-based healthcare company delivering virtual primary care services enhanced by an AI-powered diagnostic and intake tool. Their core technology leverages a large dataset of medical information to provide preliminary assessments and insights to clinicians, facilitating more efficient and data-driven patient consultations. K Health targets health systems and individuals seeking accessible, 24/7 primary care that integrates with existing in-person care networks to address access and continuity challenges.

Liquid AI

Cambridge, United States

Liquid AI develops efficient foundation models based on liquid neural networks, a new AI architecture from MIT research.

Covariant

Emeryville, United States

Covariant develops AI-powered robotic automation solutions for warehouse and logistics environments. Their core technology is the Robotics Foundation Model (RFM-1), a universal AI trained on a large dataset of warehouse robotics data, enabling robots to handle a wide variety of SKUs without specialized programming. Covariant targets large retailers and logistics providers seeking to improve operational flexibility and address labor challenges through adaptable, AI-driven automation.

Paige AI

New York, United States

Paige AI develops AI-powered tools for pathology, focusing on improving cancer diagnosis and biomarker discovery. Their core technology leverages large foundation models trained on vast pathology datasets to assist pathologists in detecting cancer, subtyping tumors, and identifying molecular markers directly from tissue samples. Paige targets pathology labs and healthcare providers, aiming to increase diagnostic accuracy and efficiency while alleviating workload pressures.

Imbue

San Francisco, United States

Imbue develops AI agents focused on software development, offering tools to improve the reliability and collaborative aspects of AI-assisted coding. Their primary product, Sculptor, is a user interface and containerization system for running and debugging multiple instances of Anthropic’s Claude Code model in parallel. This targets software developers seeking to integrate and effectively utilize large language models for coding tasks, providing a platform for safe experimentation and issue detection.

H Company

Paris, France

aiming for informative detail and professionalism: H Company builds autonomous AI agents powered by its proprietary Large Language Model, “Pithia,” designed for complex task execution and long-term memory retention. The company differentiates itself through a focus on agent reliability and interpretability, utilizing techniques like reinforcement learning from human feedback and formal verification to minimize "hallucinations" and ensure predictable behavior. Backed by $220M in funding, H Company targets enterprise automation of knowledge work, and recently demonstrated successful pilot programs with clients in the financial services and legal sectors. Key improvements & explanations of choices: Specificity: Instead of just "frontier AI agents," we mention the LLM name ("Pithia") and the type of tasks they're designed for. AI Capabilities Highlighted: We move beyond just saying they have AI capabilities and explain how they achieve them (RLHF, formal verification) and what problem those techniques solve (hallucinations, predictability). Target Market & Achievements: We move beyond just "automation" and specify what kind of automation (knowledge work) and provide evidence of traction (pilot programs, specific sectors). Removed Redundancy: The founding team information is good, but doesn't need to be in the core description. It can be added elsewhere (e.g., "About Us" section). Professional Tone: Avoids overly promotional language ("frontier").

Imply

San Francisco, United States

Imply develops the Observability Warehouse, a data layer built on Apache Druid, designed to decouple and consolidate observability and security data. Their core product enables organizations to ingest data once and utilize it across various analytical tools, including direct querying and integration with large language models like Claude and ChatGPT. Imply targets organizations struggling with data silos and high costs associated with traditional, tightly-coupled observability stacks, offering a more flexible and cost-effective solution.

01.AI

Beijing, China

01.AI develops and releases open-weight large language models (LLMs), most notably the Yi series, with a focus on strong multilingual capabilities and efficient deployment. Their flagship model, Yi-Lightning, utilizes a Mixture-of-Experts (MoE) architecture to achieve state-of-the-art performance, and the company emphasizes open-source accessibility for developers and researchers. Founded by Kai-Fu Lee, 01.AI aims to drive innovation in the "AI 2.0" era by providing foundational models and fostering a robust ecosystem around their technology.

Sakana AI

Tokyo, Japan

Sakana AI is developing next-generation foundation models inspired by principles of neuroscience and natural intelligence. The Tokyo-based research lab is focused on building large language models intended to address specific needs within the Japanese market, with a stated goal of democratizing AI access domestically. Founded by former Google researchers, Sakana AI aims to differentiate its approach through biologically-inspired AI architectures, though specific model names or performance benchmarks are not yet publicly available.

BioAge Labs

Richmond, United States

BioAge Labs is a US-based biotechnology company focused on developing therapeutics for cardiometabolic diseases by targeting the biological processes of aging. Their core technology is a human-first discovery platform that leverages decades of proprietary multi-omics data to identify and validate novel aging-related drug targets. Currently, BioAge is advancing BGE-102, a program targeting chronic inflammation, and a preclinical apelin receptor agonist with potential for significant weight loss, both aimed at addressing age-related metabolic dysfunction.

Primer AI

San Francisco, United States

Primer AI develops automated intelligence analysis platforms for the U.S. defense and intelligence communities. Their core technology utilizes advanced Natural Language Processing (NLP) – specifically large language models and machine reading – to rapidly synthesize insights from vast quantities of text data. This enables analysts to accelerate threat detection, monitor global events, and improve decision-making with greater speed and scale than traditional methods.

Perplexity AI

San Francisco, United States

Perplexity AI develops a conversational search engine powered by Large Language Models (LLMs), offering direct answers with cited sources rather than lists of links. Their core innovation lies in a proprietary blend of retrieval-augmented generation (RAG) and a focus on academic research papers, enabling more factual and nuanced responses than traditional search. Targeting information workers, researchers, and curious learners, Perplexity AI has gained traction for its "Copilot" feature – an AI-powered research assistant – and recently secured $52 million in Series B funding led by IVP and Nat Friedman.

Character.AI

Palo Alto, United States

Character.AI develops neural language model-based chatbots leveraging a proprietary large language model trained for engaging, personality-driven conversations. Their platform allows users to create and interact with AI characters – including options like a historical figure or fictional persona – and features tools for character creation and scene building. Since launching in September 2022, Character.AI has rapidly gained popularity, reaching over 1 million daily active users and demonstrating strong user retention through its focus on emotionally resonant and creative AI interactions.

Brainly

Krakow, Poland

Brainly is a Poland-based EdTech company offering an AI-powered learning platform for students. Their core product is an AI Tutor that provides instant homework help, generates personalized study materials from user inputs, and offers adaptive test preparation. Brainly targets students seeking supplemental academic support and aims to improve learning outcomes through accessible, AI-driven educational resources.

Magic

San Francisco, United States

Magic is an AI company developing frontier-scale code models designed to automate software engineering and AI research. Their core technology focuses on ultra-long context language models, leveraging 8,000 H100s to improve model performance and address AI alignment challenges. Targeting the advanced AI research community and software development organizations, Magic aims to accelerate progress towards safe Artificial General Intelligence (AGI) through automated code generation and model improvement.

Jasper

Austin, United States

Jasper AI provides an AI-powered content automation platform for marketing teams, streamlining the entire content lifecycle from planning to execution. Their core technology utilizes intelligent AI Agents and a contextual knowledge layer (“Jasper IQ”) to generate on-brand content and automate workflows via tools like Jasper Grid and Jasper Studio. Jasper targets enterprise marketing organizations seeking to increase content velocity, maintain brand consistency, and scale content creation efficiently.

Hasura

Bangalore, India

Hasura is a data access company specializing in tools for building and scaling data-intensive applications, particularly those leveraging AI. Their core product, PromptQL, is a data delivery network designed to provide fast, accurate, and contextually relevant data to large language models and other AI systems. Targeting enterprise data leaders and rapidly growing AI-native companies, Hasura aims to simplify data integration and accelerate the development of reliable AI-powered experiences.

Snorkel AI

Redwood City, United States

Snorkel AI is a data-centric AI platform specializing in the programmatic development of high-quality training datasets. Their core technology utilizes programmatic labeling techniques to accelerate data curation and improve model accuracy, particularly for large language models and enterprise AI applications. Snorkel AI targets organizations requiring specialized, rapidly-developed datasets to optimize performance and reduce the time-to-deployment of their AI initiatives.

Writer

San Francisco, United States

Writer provides full-stack generative AI platform for enterprises with Palmyra LLMs, AI guardrails, and Knowledge Graph.

Poolside

Paris, France

Poolside is a French foundation model company developing AI agents specifically for software development and deployment within enterprise environments. Their core technology focuses on building and operationalizing large language models to enhance developer productivity and automate coding tasks. Unlike many AI providers, Poolside prioritizes data security and control by offering on-premise, VPC, or workstation deployment options, targeting organizations with stringent compliance and security requirements.

Supabase

San Francisco, United States

Supabase is a backend-as-a-service platform built on top of PostgreSQL, providing developers with a comprehensive suite of tools including authentication, real-time subscriptions, and serverless functions. A key differentiator is its integrated vector database capabilities, enabling efficient storage, indexing, and search of vector embeddings for AI and machine learning applications. Supabase targets developers building full-stack applications requiring a scalable database solution with native AI functionality, offering an open-source alternative to platforms like Firebase.

Cohere Health

Boston, United States

Cohere Health applies generative AI to automate and streamline the prior authorization process for healthcare payers and providers. Their platform utilizes clinically-trained large language models to reduce administrative burden and improve collaboration around utilization management. Cohere Health targets health plans and provider groups seeking to improve efficiency, reduce costs, and accelerate patient access to care by optimizing payment integrity.

Neon

San Francisco, United States

Neon is a serverless Postgres database provider offering autoscaling and pay-as-you-go pricing for application development. Their platform uniquely integrates pgvector, enabling efficient storage and retrieval for AI-powered applications utilizing vector embeddings. Neon targets developers building scalable applications, particularly those leveraging AI and machine learning, by simplifying database operations and reducing infrastructure management overhead.

Yellow.ai

Bangalore, India

Yellow.ai is an Indian enterprise AI platform specializing in the development and deployment of autonomous agents for customer experience (CX) and employee experience (EX) automation. Their platform leverages a suite of 15+ large language models to deliver scalable, high-quality conversational AI solutions. Yellow.ai targets large enterprises seeking to reduce operational costs and improve service efficiency through the automation of routine interactions across both customer and employee channels.

Upstage

Seoul, South Korea

Upstage is a South Korean AI company specializing in document processing and large language models (LLMs) for enterprise applications. Their core offering is a suite of LLMs – including the Solar model family – and accompanying document AI tools that convert and extract structured data from complex documents like invoices, contracts, and clinical records. Upstage targets industries requiring high accuracy and reliability – such as insurance and healthcare – by automating workflows and enabling LLM-powered insights from unstructured data.

Merantix

Berlin, Germany

Merantix is a German AI venture studio that develops and launches independent AI-driven companies. They focus on applied AI research and development, building solutions across multiple industries rather than offering a single product. Merantix targets opportunities where AI can deliver significant impact, functioning as both a creator and investor in these new ventures.

You.com

Palo Alto, United States

You.com develops a search infrastructure platform leveraging large language models to provide real-time, accurate search results for enterprise applications. Their core offering is a Search API and customizable vertical indexes designed for Retrieval-Augmented Generation (RAG) and agentic AI workflows, emphasizing data freshness and minimizing hallucinations. You.com differentiates itself through a focus on delivering citation-backed, structured data with proven reliability and performance, positioning them as a key provider for businesses building next-generation AI agents and applications.

Contextual AI

Mountain View, United States

Contextual AI builds RAG-native language models designed specifically for enterprise knowledge work and document understanding.

Fireworks AI

Redwood City, United States

Fireworks AI is a US-based platform specializing in accelerated inference for open-source generative AI models, including Large Language and image models. Their core offering is a cloud-based inference platform optimized for speed, cost, and quality, enabling users to both utilize pre-trained models and fine-tune/deploy custom models. They target enterprises seeking to build and scale generative AI applications like chatbots, knowledge base tools, and personalized recommendation systems without the infrastructure burden.

Rasa

Berlin, Germany

Rasa is a German-based company providing an open-source conversational AI platform for building scalable and customized AI agents. Their platform extends Large Language Models (LLMs) with proprietary business logic, enabling enterprises to deploy reliable, high-performance agents across multiple channels. Rasa primarily targets enterprise teams seeking full control and customization over their conversational AI, particularly for complex use cases like customer service and support where scalability and integration with existing systems are critical.

Weaviate

Amsterdam, Netherlands

Weaviate is an open-source vector database that enables developers to build AI-native applications with semantic search capabilities. The platform simplifies AI infrastructure by handling vector embeddings, ranking, and auto-scaling, allowing users to connect custom or pre-built machine learning models without complex data pipelines. Weaviate targets developers and enterprises seeking a scalable, secure, and vendor-agnostic solution for knowledge graphs, recommendation engines, and other AI-powered applications.

Fiddler AI

Palo Alto, United States

Fiddler AI provides an AI observability platform specializing in the monitoring and analysis of machine learning models, including those powering large language models and AI agents. Their technology focuses on explainability and performance management, identifying and resolving issues like model drift and bias in production environments. Fiddler targets enterprises deploying AI at scale who require robust monitoring and governance to ensure reliable and responsible AI applications.

Prefect

Washington DC, United States

Prefect is a workflow orchestration platform that enables reliable automation of data, machine learning, and agent-based workflows. Their core technology centers on a Python-native workflow engine allowing users to run existing code without requiring specialized languages or rigid DAG structures. Prefect targets data science and engineering teams seeking a unified control plane to manage and scale complex, context-aware workflows – including those leveraging large language models – from experimentation to production.

Codeium

Mountain View, United States

Codeium develops Windsurf, an AI-native integrated development environment (IDE) and coding assistant. Utilizing a proprietary AI model called Cascade, Windsurf provides code completion and contextual awareness across 70+ programming languages. Targeting professional developers and enterprise teams, Codeium aims to enhance coding efficiency and maintain developer workflow through its AI-powered tools.

Unstructured

San Francisco, United States

Unstructured is a US-based company specializing in data transformation for generative AI applications. They offer an open-source and enterprise solution that extracts and structures data from a variety of unstructured documents, preparing it for use in Retrieval-Augmented Generation (RAG) pipelines. Targeting large enterprises – including 82% of the Fortune 1000 – Unstructured differentiates itself through a focus on data security, compliance, and continuous data processing for LLM readiness.

Quizlet

San Francisco, United States

Quizlet is an EdTech company that leverages AI-powered adaptive learning technology within its platform of digital flashcards and study tools. Their core offering, Q-Chat, functions as an AI tutor providing personalized explanations and practice, while the platform dynamically adjusts study materials based on individual student performance. Quizlet primarily serves students and teachers across all educational levels, aiming to improve comprehension and learning outcomes through accessible, AI-driven study aids.

Arize AI

Berkeley, United States

Arize AI provides a unified platform for monitoring and evaluating Large Language Model (LLM) applications, from development through production deployment. Their technology focuses on LLM observability and agent evaluation, offering tools to analyze performance, identify issues, and close the feedback loop between development and real-world data. Arize AI targets organizations building and scaling AI applications, enabling data-driven iteration and improved reliability at scale.

Reka AI

San Francisco, United States

Reka AI builds multimodal foundation models that can understand text, images, video, and audio in a unified architecture.

Essential AI

San Francisco, United States

Essential AI, founded by former Google researchers, builds enterprise-focused foundation models optimized for business workflows.

Evisort

San Francisco, United States

Evisort is a contract lifecycle management (CLM) platform that utilizes a proprietary, purpose-built large language model (LLM) specifically trained on contract data. This AI-native approach enables automated contract analysis, risk identification, and extraction of financial insights for businesses. Targeting enterprises and growing companies, Evisort differentiates itself from competitors by offering a fully integrated AI solution – rather than relying on generic, third-party AI – to improve contract accuracy, security, and control.

Krutrim

Bangalore, India

Krutrim, founded by Ola's Bhavish Aggarwal, builds India-first AI with multilingual LLMs supporting 22 Indian languages.

AI inside

Tokyo, Japan

AI inside is a Japanese technology company specializing in data automation through advanced Optical Character Recognition (OCR) and Retrieval-Augmented Generation (RAG). Their core product is a platform enabling businesses to build custom AI agents for automating data-intensive workflows, leveraging a proprietary, Japanese-language focused Large Language Model and on-premise edge computing for enhanced security. They target enterprises seeking to digitize and automate document processing, particularly those requiring high accuracy and data privacy within a secure, localized infrastructure.

Predibase

San Francisco, United States

Predibase is an enterprise AI platform specializing in the operationalization of Large Language Model (LLM) powered agents. Their core technology, Rubrik Agent Cloud, provides monitoring, governance, and a unique “rewind” capability allowing businesses to revert unintended agent actions without data loss or operational disruption. Predibase targets enterprises seeking to deploy and manage AI agents at scale while maintaining control, auditability, and risk mitigation.

Haystack (deepset)

Berlin, Germany

deepset develops Haystack, an open-source framework and accompanying platform for building and deploying Retrieval-Augmented Generation (RAG) applications and AI agents. Their technology enables enterprises to connect Large Language Models (LLMs) to proprietary data sources, improving accuracy and reducing the risk of misinformation in AI-driven solutions. deepset targets businesses seeking to rapidly implement and scale custom AI applications with a focus on data control, transparency, and production readiness.

Phrasee

London, United Kingdom

Phrasee offers a SaaS platform, Jacquard, that utilizes a proprietary neural network trained on over 60 billion data points to generate and optimize performance-driven marketing copy for channels like email, SMS, and push notifications. Unlike general-purpose LLMs, Jacquard is purpose-built for brand messaging at scale, focusing on statistically predicting and improving click-through rates. The company targets enterprise marketing teams seeking to overcome content bottlenecks, reduce operational costs, and improve Customer Lifetime Value through data-driven, brand-compliant messaging.

TruEra

Redwood City, United States

TruEra provides AI observability solutions for monitoring, testing, and managing the quality of machine learning models throughout their lifecycle. Their platform focuses on both traditional predictive AI and Large Language Models (LLMs), offering specialized observability for each. TruEra targets organizations implementing MLOps and LLMOps seeking to mitigate risks and ensure reliable performance of their AI deployments, and is now a part of Snowflake.

Robust Intelligence

San Francisco, United States

Robust Intelligence, now part of Cisco, develops AI security solutions focused on identifying vulnerabilities in and protecting against attacks on machine learning models. Their core technology utilizes algorithmic red teaming – including a “Tree of Attacks” methodology – to proactively test and secure LLMs like GPT-4 and Llama-2 against jailbreaks and harmful output generation. This platform serves organizations deploying AI applications who require robust security testing and mitigation to confidently advance their AI initiatives, and is foundational to Cisco’s AI Defense and Foundation AI offerings.

Sarvam AI

Bangalore, India

Sarvam AI develops a full-stack generative AI platform focused on creating a sovereign AI ecosystem for India. Their core offering is a multilingual large language model supporting 11 Indian languages, enabling the development of AI agents and voice applications. Sarvam AI targets government entities, enterprises, and developers seeking localized AI solutions tailored to the Indian market and regulatory landscape.

Latitude

Seattle, United States

Latitude is a US-based AI company specializing in generative AI for interactive entertainment. They develop and deploy large language models – most notably their proprietary model used in AI Dungeon – to power dynamic, open-world text-based adventure games. Latitude targets a niche market of players and developers seeking highly customizable and emergent narrative experiences beyond traditional game structures.

LangChain

San Francisco, United States

LangChain is a US-based developer tools company providing an open-source framework and engineering platform for building applications powered by Large Language Models (LLMs). Their core offering centers on “chains,” “agents,” and memory modules that enable developers to construct complex LLM-based workflows and applications with enhanced observability through their LangSmith tracing tool. LangChain targets AI developers and engineering teams seeking to rapidly prototype, deploy, and monitor reliable LLM applications without vendor lock-in, offering both pre-built architectures and low-level customization options.

Skelter Labs

Seoul, South Korea

Skelter Labs is a South Korean AI company specializing in Korean-language conversational AI and Natural Language Understanding (NLU) technology, originating as a spin-off from Google. Their core product focuses on advanced generative AI models tailored for nuanced Korean linguistic contexts. Skelter Labs targets businesses requiring highly accurate and culturally-relevant Korean language processing, particularly in applications like chatbots, virtual assistants, and automated customer service.

Anyword

New York, United States

Anyword is a US-based AI platform that optimizes marketing copy through predictive performance scoring. Utilizing a proprietary, LLM-agnostic engine and a large dataset of A/B tested results, Anyword predicts the effectiveness of content variations with 82% accuracy – significantly outperforming standard large language models. The platform targets marketing teams and enterprises seeking to improve campaign ROI and maintain data privacy through secure, compliant AI-driven content generation and optimization.

Bioptimus

Paris, France

Bioptimus is building universal AI foundation models for biology. Their models understand biological systems to accelerate drug discovery and healthcare innovation.

Qwak

Tel Aviv, Israel

Qwak, now operating as JFrog ML, provides a comprehensive MLOps platform for the end-to-end lifecycle of machine learning models, including Generative AI and Large Language Models. Their platform centralizes model development, deployment, and monitoring with features like automated training, scalable inference options, and dedicated LLM observability tools – including prompt management and workflow tracing. Qwak targets data science and machine learning engineering teams seeking to accelerate and reliably scale AI applications from prototype to production.

Black Forest Labs develops generative AI foundation models focused on image generation and editing. Their core product is FLUX, a family of models—including FLUX.2 and FLUX Kontext—capable of producing photorealistic 4MP images with multi-reference control, accessible via API. Founded in 2024, the company targets developers and businesses requiring high-performance, production-ready image AI capabilities.

Cleanlab

San Francisco, United States

Cleanlab provides a post-hoc AI safety layer that identifies and remediates incorrect or unsafe outputs from any deployed AI agent, including those powered by large language models. Their core technology utilizes a proprietary confidence-based approach to detect label errors and predict potentially harmful responses without requiring retraining or modifications to existing AI infrastructure. Cleanlab targets enterprises prioritizing AI safety, compliance, and trustworthiness, offering a deployable solution for maintaining quality control over generative AI and other AI-driven applications.

iGenius

Milan, Italy

iGenius, now operating under the brand Domyn, provides a complete AI platform for regulated industries seeking full ownership and control of their data and models. Their core technology is an orchestration center and knowledge graph enabling the building, deployment, and governance of large language models (LLMs) with a focus on privacy, auditability, and computational efficiency. Domyn targets enterprises in highly regulated sectors requiring sovereign AI solutions and independence from external AI providers, offering a pathway to manage the entire AI lifecycle internally.

Wombo

Toronto, Canada

Wombo develops consumer AI applications and has expanded into distributed computing infrastructure. Their w.ai platform allows users to passively earn income by securely contributing their device’s idle processing power to a decentralized network. This network supports AI model training and aims to democratize access to computational resources for the development of artificial intelligence, backed by NVIDIA.

LightOn

Paris, France

LightOn develops on-premise Retrieval-Augmented Generation (RAG) solutions powered by optical co-processors, enabling enterprises to securely query and analyze unstructured data. Their “Paradigm” platform delivers LLM-powered search and reasoning capabilities directly within a company’s infrastructure, addressing data privacy and compliance needs like GDPR, SOC 2, and HIPAA. LightOn targets organizations—including those in aerospace, government, and digital marketing—requiring secure, customizable AI solutions for knowledge management and content creation.

Wrtn

Seoul, South Korea

Wrtn is a Korean AI platform offering unlimited free access to GPT models, becoming one of Korea's most popular consumer AI apps.

Silo AI

Helsinki, Finland

Silo AI is a Finnish AI lab specializing in the development and deployment of custom AI models and large language models for enterprise clients. The company focuses on optimizing AI solutions for high-performance compute platforms, leveraging a team of over 300 AI scientists and researchers. Through a combination of applied research and a significant, publicly-funded initiative (“Compute to Impact”), Silo AI aims to accelerate AI innovation and provide a full-stack solution – from model development to scalable deployment – for businesses seeking a competitive advantage.

GigaSpaces

Tel Aviv, Israel

GigaSpaces provides an in-memory computing platform that enables enterprises to perform real-time AI-powered analytics directly on their existing structured data. Their core product utilizes Retrieval-Augmented Generation (RAG) to facilitate natural language querying and contextual insights without data replication or ETL processes. Targeting businesses seeking to democratize data access and accelerate decision-making, GigaSpaces offers a rapidly deployable SaaS solution for operational data analysis and strategic planning.

Conjecture

London, United Kingdom

Conjecture is a UK-based AI research company developing a novel AI architecture focused on verifiable safety and control. Their core technology centers on building AI systems with inherent transparency and predictability, moving beyond traditional black-box approaches. This positions Conjecture to serve organizations and researchers prioritizing robust AI safety and alignment, particularly as advanced AI capabilities continue to develop.

Exafunction

Mountain View, United States

Exafunction develops Windsurf, an AI-native integrated development environment (IDE) and coding assistant powered by large language models. Windsurf aims to improve developer productivity by providing contextual code completion, remembering codebase specifics, and maintaining workflow continuity. The company targets professional software developers and enterprises seeking to accelerate development cycles and improve code quality through AI-assisted tools.

Botpress

Quebec City, Canada

Botpress is a Canadian company offering an open-source platform for developing and deploying AI agents. Their platform distinguishes itself through a no-code/low-code studio environment and extensive integration capabilities, enabling businesses to automate a wide range of workflows. Botpress targets developers and organizations seeking a flexible, self-hosted solution for building conversational AI applications powered by large language models (LLMs).

Twelve Labs

San Francisco, United States

Twelve Labs is a US-based AI company specializing in comprehensive video understanding through its foundational model, Marengo. Marengo delivers industry-leading performance in semantic video search and analysis, exceeding the capabilities of existing cloud-based and open-source alternatives. The company targets enterprise clients requiring advanced video intelligence for applications like content discovery, asset management, and automated video workflows.

Lakera

Zurich, Switzerland

Lakera is a Swiss AI security company specializing in protecting Large Language Model (LLM) applications from emerging threats like prompt injection. Their core technology leverages a continuously learning threat model, informed by data from their widely-used “Gandalf” cybersecurity game and a leading AI red team, to provide real-time exploit detection. Lakera targets enterprise organizations adopting Generative AI, offering a proactive security layer to accelerate and safeguard their AI initiatives.

Chroma

San Francisco, United States

Chroma provides an open-source, embeddable vector database designed for building applications powered by Large Language Models (LLMs). The company’s core technology focuses on efficient semantic search and retrieval of data via vector embeddings, enabling developers to add memory and context to AI applications. Chroma targets developers and organizations seeking a scalable, customizable, and open-source alternative to proprietary vector database solutions for LLM-powered applications like chatbots, knowledge bases, and semantic search tools.

Spellbook

Toronto, Canada

Spellbook is a Canadian Legal AI company that streamlines contract workflows for transactional lawyers. Utilizing large language models, including GPT-4o, Spellbook’s software integrates directly with Microsoft Word to automate contract drafting, redlining, and risk assessment. The platform differentiates itself by enabling users to leverage existing precedents while also benchmarking against over 2,000 common legal standards.

Neptune.ai

Warsaw, Poland

Neptune.ai provides experiment tracking and model registry software specializing in the monitoring of large-scale foundation models. Their platform uniquely focuses on visualizing and debugging per-layer metrics – including losses, gradients, and activations – at scale, enabling faster identification of training instabilities. Targeting AI research and infrastructure teams working with models exceeding billions of parameters, Neptune.ai offers both cloud and on-premise deployment options for comprehensive model monitoring.

Layer 6

Toronto, Canada

Layer6 is the AI center of excellence for TD Bank Group, focused on developing and deploying advanced AI solutions for the financial services industry. Their core technology centers on generative AI, specifically Retrieval-Augmented Generation (RAG) and Text-to-SQL models – exemplified by the TD Securities AI Virtual Assistant – alongside novel research in automated causal inference like their CausalPFN model. Layer6 uniquely translates cutting-edge AI research into impactful applications for over 27 million customers, providing data-driven insights and personalized financial services while contributing to the Canadian AI ecosystem.

Surge AI

San Francisco, United States

Surge AI is a US-based data labeling company specializing in high-quality training data for Reinforcement Learning from Human Feedback (RLHF) and Large Language Models (LLMs). Their core offering is a managed data labeling platform focused on complex annotation tasks like preference labeling and reward modeling, critical for aligning AI behavior. Surge AI targets AI developers and research teams building and refining generative AI applications requiring nuanced human input for optimal performance.

Convergence AI develops autonomous AI agents powered by a novel, scalable reinforcement learning framework built upon large language models. Their technology focuses on enabling agents to solve complex tasks with minimal human intervention, particularly within enterprise workflows. Notably, they recently achieved significant performance benchmarks in multi-agent collaboration challenges, demonstrating advanced AI coordination capabilities.

Cbot

Istanbul, Turkey

Cbot is a Turkish AI company specializing in generative AI-powered conversational solutions for enterprise contact centers. Their core product is an intelligent virtual assistant leveraging large language models – specifically a GPT-based “Human + AI” hybrid – to automate customer service interactions and seamlessly escalate complex issues to live agents. Cbot targets businesses seeking to improve customer experience and operational efficiency through 24/7 personalized support in Turkish and other languages.

Elicit

San Francisco, United States

Elicit is a research AI company that leverages large language models to automate the analysis of scientific and clinical literature. Their core product is an AI research assistant capable of searching, summarizing, and extracting key data points from a database exceeding 138 million academic papers and clinical trials. Elicit primarily serves the academic and industry research communities, offering a solution to accelerate literature review and knowledge discovery.

Saidot

Helsinki, Finland

Saidot is a Finnish SaaS provider specializing in AI governance solutions. Their platform enables organizations to document, assess, and monitor AI systems to ensure responsible development and adherence to the forthcoming EU AI Act. Saidot targets businesses and public sector entities requiring demonstrable AI compliance and risk management capabilities within the European regulatory landscape.

Meta AI

Menlo Park, United States

Meta AI (formerly FAIR) is Meta's AI research division. Creator of LLaMA, PyTorch, and leading research in computer vision, NLP, and embodied AI.

Alibaba DAMO Academy

Hangzhou, China

DAMO Academy is Alibaba's global research initiative covering AI, computer vision, and autonomous systems. Creator of Qwen models.

Tencent AI Lab

Shenzhen, China

Tencent AI Lab is a research division of Tencent focused on advancing artificial intelligence capabilities across multiple modalities including natural language processing and computer vision. Their primary development is Hunyuan, a large language model intended for integration into Tencent’s extensive product ecosystem. The lab targets internal applications within Tencent’s services, as well as potential commercialization of AI solutions within the Chinese market.

ByteDance AI Lab

Beijing, China

ByteDance AI Lab focuses on developing core AI technologies powering the TikTok platform and beyond, including the recommendation engine “People You May Know” and computer vision models for content moderation and effects like green screen. Their key innovations lie in large-scale reinforcement learning for personalized recommendations, and they are a leading developer of Chinese Large Language Models, including the series known as “Qwen,” released as open-source models. With over 300 research papers published in top AI conferences, ByteDance AI Lab is a significant contributor to advancements in areas like text-to-image generation and video understanding.

Naver AI

Seongnam, South Korea

Naver AI Lab is the artificial intelligence research and development division of NAVER Corporation, focused on advancing core AI technologies for its search engine and broader services. Their primary development is HyperCLOVA, a large language model (LLM) powering features like query understanding, content generation, and translation. Targeting the Korean language market specifically, Naver AI Lab differentiates itself by building LLMs optimized for the nuances and complexities of Korean, a historically underserved language in AI development.

Kakao Brain

Seongnam, South Korea

Kakao Brain is a leading AI research lab specializing in large language and vision models, most notably developing Korea’s first native large language model, KoGPT. Their innovations include hyper-realistic image generation models like DALL-E-inspired “MinDALL-E” and advancements in AI-powered services integrated within Kakao’s widely-used messaging app and portal, including AI-driven chatbots and image search. Kakao Brain recently open-sourced several of their models and datasets, contributing to the growth of the AI ecosystem in Korea and beyond, and was integrated into Kakao Enterprise in 2023 to focus on business applications.

Redwood Research

Berkeley, United States

Redwood Research is a US-based AI safety nonprofit focused on mitigating catastrophic risks from advanced AI systems. Their core research centers on “AI control” protocols – techniques for reliably monitoring and preventing subversion by potentially deceptive large language models, even when those models intentionally conceal misaligned intentions. They serve as a critical resource for both governments and leading AI developers like Google DeepMind and Anthropic, providing expertise and methodologies for assessing and mitigating AI safety risks.

Apollo Research

London, United Kingdom

Apollo Research is a UK-based AI safety and alignment company specializing in the detection of deceptive behavior in advanced AI systems, particularly large language model agents. They develop and implement novel AI model evaluations focused on “scheming” – covertly pursuing misaligned objectives – and provide technical expertise to governments and international organizations on AI governance and regulation. Their core offering is third-party evaluation of frontier AI models, alongside consultancy services for responsible AI development frameworks and policy guidance.

Center for AI Safety

San Francisco, United States

The Center for AI Safety is a US-based nonprofit focused on mitigating potentially catastrophic risks from advanced artificial intelligence. They conduct and fund research into AI safety, with a particular emphasis on identifying and addressing vulnerabilities in increasingly capable AI systems – offering resources like a dedicated compute cluster to support this work. Their primary audience includes AI researchers, policymakers, and stakeholders concerned with the responsible development and deployment of powerful AI technologies.

AI Fund

Palo Alto, United States

AI Fund is a venture studio that actively co-founds AI-driven companies alongside entrepreneurs and corporate partners. Leveraging over $370 million in backing and the expertise of Andrew Ng, they provide pre-seed funding, technical talent, and a company-building methodology to rapidly launch ventures. AI Fund focuses on applying AI to solve complex problems across industries like maritime, finance, and relationship management, partnering with those possessing deep domain expertise.

Intercom Fin

San Francisco, United States

Intercom Fin is a SaaS platform offering an AI agent for customer service automation. Utilizing a GPT-4 foundation and a proprietary “Fin Flywheel” continuous improvement loop, the platform is trained on a company’s specific knowledge base and policies to resolve complex customer queries across multiple channels. Intercom Fin targets businesses seeking to improve customer support efficiency and consistency through AI-powered automation and performance analytics.

Haptik

Mumbai, India

Jio Haptik develops conversational AI agents for enterprise customer service, enabling automated interactions across voice and digital channels. Their core offering is a no/low-code platform allowing businesses to build and deploy custom agents powered by large language models like GPT, Llama, and Claude. Targeting businesses requiring scalable, multilingual support, Jio Haptik differentiates itself by offering a platform for rapid deployment and integration with existing communication infrastructure.

Technology Innovation Institute

Abu Dhabi, United Arab Emirates

Technology Innovation Institute (TII) is a UAE-based research center focused on advanced technology development, with a core competency in artificial intelligence. TII is the creator of the open-source Falcon series of large language models, notable for their performance and accessibility. As part of the Abu Dhabi government’s research ecosystem, TII aims to advance scientific knowledge and deliver impactful technologies through both independent research and international collaboration.

Khan Academy

Mountain View, United States

Khan Academy is a non-profit educational organization leveraging OpenAI’s GPT-4 to deliver personalized learning experiences through its AI tutor, Khanmigo. This AI-powered tool provides students with individualized guidance, practice problems, and feedback across a range of subjects, supplementing traditional learning resources. Khan Academy primarily serves students and educators globally, offering a free, accessible alternative and enhancement to conventional educational methods.

Notion AI

San Francisco, United States

Notion AI is a productivity software company that embeds a large language model directly within its existing all-in-one workspace platform. This integration provides AI-powered features like writing assistance, content summarization, and intelligent search to streamline workflows. Targeting knowledge workers and teams, Notion AI differentiates itself by offering AI functionality within a comprehensive note-taking, project management, and wiki system, rather than as a standalone tool.

Huawei AI

Shenzhen, China

Huawei develops the Ascend series of AI chips – including the Ascend 910 and NPU-based modules – designed to accelerate machine learning workloads for edge and cloud deployments. Their AI capabilities are demonstrated through the Pangu large language model series, which includes models like Pangu-α and Pangu-Weather, showcasing advancements in natural language processing and weather forecasting accuracy. Primarily serving the telecommunications, manufacturing, and financial services sectors, Huawei AI has achieved notable deployments of its solutions in smart city initiatives and industrial automation across China and internationally.

DeepSeek

Hangzhou, China

DeepSeek is a China-based AI company developing and releasing open-source large language models (LLMs) focused on coding and general reasoning capabilities. Their product line includes DeepSeek-LLM, DeepSeek-Coder, and the recently released DeepSeek-MoE, utilizing a Mixture-of-Experts architecture for enhanced performance. DeepSeek targets developers and researchers by providing both open-source models and commercial API access to their advanced LLMs.

Stats Perform

Chicago, United States

Stats Perform is a sports AI company that provides detailed, real-time data and analytics to the global sports industry. Their core product, Opta, is a comprehensive sports data engine powering 8 proprietary AI models used for predictive analytics and performance insights. Stats Perform serves professional sports teams, broadcasters, media outlets, and betting operators, enabling enhanced content creation, strategic decision-making, and improved fan engagement.

Be My Eyes AI

Copenhagen, Denmark

Be My Eyes AI provides accessibility solutions for individuals who are blind or have low vision by leveraging both human volunteers and AI-powered visual assistance. Their core offering integrates GPT-4 Vision to deliver instant image descriptions via a mobile app, supplementing a pre-existing network of over 9.6 million volunteers providing live visual support. Beyond individual users, Be My Eyes also offers a business solution enabling companies to enhance customer service and workplace accessibility through AI-driven video assistance.

Feast

San Francisco, United States

Feast provides an open-source feature store designed to streamline the machine learning lifecycle. Their platform enables data scientists and ML engineers to define, manage, and serve features consistently across model training and real-time inference, addressing a critical need for operationalizing ML models at scale. Feast targets organizations building and deploying machine learning applications – particularly those leveraging large language models and real-time personalization – by providing a centralized repository for feature data and supporting integrations with popular MLOps tools like Ray and Kubeflow.

Ray

San Francisco, United States

Ray (Anyscale) provides a unified, open-source distributed computing framework designed to simplify the development and deployment of AI and machine learning applications. Their core technology enables developers to scale Python-based AI workloads – including training and serving for models like LLMs and computer vision applications – across diverse infrastructure, from CPUs to GPUs. Ray targets organizations facing challenges with scaling AI initiatives and optimizing resource utilization, offering a solution to overcome the “AI Complexity Wall” and accelerate time to production.

LG AI Research

Seoul, South Korea

LG AI Research focuses on developing and deploying large language models, with its flagship product being EXAONE, a 1.3 trillion parameter LLM trained on a massive dataset of text and code. The company is a key driver of Korea’s national AI strategy, spearheading the country’s sovereign AI initiative and focusing on advancements in generative AI and multimodal models. LG AI Research aims to apply these technologies across LG’s diverse business portfolio – including appliances, electronics, and business solutions – and is positioned to serve enterprise clients seeking customized AI solutions.

TensorRT

Santa Clara, United States

NVIDIA TensorRT is an SDK for optimizing and deploying deep learning models across a range of hardware, from data centers to edge devices. Utilizing techniques like quantization and kernel tuning, TensorRT significantly reduces inference latency and increases throughput compared to CPU-only deployments. The platform specifically targets developers working with performance-critical applications and large language models requiring efficient GPU acceleration.

CloudWalk

Guangzhou, China

CloudWalk Technology is a Chinese AI platform company specializing in human-machine collaboration operating systems and applied AI solutions. Their core technology integrates computer vision – including facial recognition and cross-camera Re-ID – with natural language processing and large models to bridge the digital and physical worlds. CloudWalk targets sectors including finance, urban management, and commercial applications, offering both platform-level AI capabilities and tailored industry solutions.

GitLab

San Francisco, United States

GitLab provides a comprehensive DevSecOps platform integrating AI-powered features throughout the software development lifecycle. Their core offering, GitLab Duo, utilizes large language models to deliver contextual code suggestions, automated vulnerability detection, and conversational AI assistance directly within the platform. Targeting professional software development teams, GitLab aims to accelerate development velocity and improve code security by embedding AI capabilities into every stage of the process.

Oracle Cloud AI

Austin, United States

Oracle Cloud AI provides a comprehensive suite of cloud-based artificial intelligence services for enterprise customers. Their core offering centers on a platform for building, training, and deploying both proprietary and open-source large language models (LLMs), alongside pre-built AI services like anomaly detection and NLP. Oracle differentiates itself by integrating these AI capabilities directly within its existing cloud applications and database services, and through partnerships offering access to models like Google Gemini, enabling businesses to leverage AI across core functions.

Slack AI

San Francisco, United States

Slack AI integrates large language models directly into its workplace messaging platform to enhance team productivity and workflow automation. Their core offering is a platform enabling collaborative work with AI agents – including integrations with Agentforce, Claude, and Google Agent Space – directly within Slack channels. This positions Slack AI as a solution for businesses seeking to leverage generative AI for tasks like content creation, coding, and strategic planning, all within their existing communication workflows.

TikTok AI

Beijing, China

TikTok AI develops and deploys machine learning algorithms to personalize content recommendations within the TikTok short-form video platform. Their core technology centers on a deep learning-based recommendation system that analyzes user interactions and video attributes to maximize engagement. Targeting a global user base of over one billion, TikTok AI differentiates itself through highly effective, rapidly adapting algorithms optimized for mobile video consumption.

BloombergGPT

New York, United States

BloombergGPT is a large language model (LLM) specifically trained on a massive dataset of financial data, including Bloomberg’s extensive news and data archives. This allows the model to perform complex natural language processing tasks tailored to the financial industry, such as sentiment analysis, entity recognition, and report generation. BloombergGPT targets financial professionals and institutions seeking to automate data-driven insights and improve efficiency in areas like investment research and risk management.

Bytedance

Beijing, China

ByteDance is a Chinese technology company developing and operating globally-reaching content platforms, most notably TikTok and Douyin. Their core AI technology centers on recommendation algorithms and machine learning models that personalize content feeds to maximize user engagement. Targeting a broad demographic of content consumers and creators, ByteDance differentiates itself through highly effective AI-driven content discovery and delivery at scale.

SDAIA

Riyadh, Saudi Arabia

The Saudi Data and AI Authority (SDAIA) is the governmental body responsible for developing and executing Saudi Arabia’s national AI strategy. SDAIA focuses on establishing a robust data and AI ecosystem through initiatives like national data platforms and AI-powered solutions for public sector applications. Its primary value proposition is to accelerate Saudi Arabia’s digital transformation and achieve the goals outlined in Vision 2030 by leveraging data and artificial intelligence.

Trillion Labs

Seoul, South Korea

Trillion Labs built Tri-70B, the largest Korean-specialized LLM at 70 billion parameters. Leading domestic AI model developer.

Doubao

Beijing, China

Doubao (豆包) is ByteDance's AI chatbot and large language model. It's one of China's most popular AI assistants with millions of users, offering conversational AI, creative writing, coding assistance, and multimodal capabilities.

EleutherAI

Remote, United States

EleutherAI is a US-based research collective focused on creating and openly releasing large language models (LLMs). Their core technology centers on training and analyzing LLMs, with a current research emphasis on eliciting and interpreting internal model knowledge (“ELK”) to improve transparency and verifiability. They uniquely serve the AI research community and developers by providing accessible, powerful open-source LLMs and tools for studying model behavior.

Cohere for AI

Toronto, Canada

Cohere for AI develops and openly releases large language models and related machine learning research. Their core technology centers on foundational LLM development, with a strong emphasis on open science and collaborative research. Cohere uniquely targets the academic, civic, and impact-focused sectors by providing free API access and fostering a broad open-source community to accelerate responsible AI development and deployment.

Nous Research

San Francisco, United States

Nous Research develops and trains high-performing, open-source large language models (LLMs) with a focus on distributed and unbiased training methodologies. Their core offering is the creation of these LLMs, made publicly available for research and commercial use. Founded in 2023, Nous Research positions itself as a key contributor to the growing US open-source AI ecosystem, aiming to provide alternatives to closed-source models.

Baichuan AI develops large language models and generative AI tools, focusing on natural language understanding, information retrieval, and reinforcement learning. Their core technology combines supervised fine-tuning with human alignment to deliver strong performance in knowledge-based question answering and text generation. As a leading Chinese AI research firm, Baichuan AI aims to democratize access to advanced AI capabilities within the region.