Connect your agent to akta.pro and build company lists like this across sectors and geos.
Private open-source ai model companies, Series A+ (round raised in 2026), ranked by total capital raised. Data compiled .
| Company | Category | Most Recent Round | Total Funding | Offering | Management | Revenue Model |
|---|---|---|---|---|---|---|
| Moonshot AI ๐จ๐ณ Haidian | Open-Source AI Model Development | Series F Jul 2026 | Funding $20.05B | Moonshot AI develops and sells the Kimi family of large language models and AI assistants. Its core products include the Kimi consumer chatbot (kimi.com), the Kimi Work desktop AI agent for enterprise productivity, and the platform.kimi.ai API giving developers programmatic access to Kimi K2.5/K2.6/K2.7 models. Open-weight model releases on Hugging Face drive ecosystem adoption that converts into paid API and enterprise contracts. | Yang Zhilin Co-founder and CEO Xinyu Zhou Co-founder Wu Yuxin Co-founder | API Consumption Kimi Chatbot Subscriptions Enterprise AI Services |
| Mistral AI ๐ซ๐ท Paris | Open-Source AI Model Development | Series D Sep 2026 | Funding $7.50B | Mistral AI develops and sells frontier large language models and a full-stack enterprise AI platform. Its product portfolio includes proprietary and open-weight foundation models (Mistral Large 3, Medium 3.5, Small 4, Codestral, Devstral, Voxtral, Mixtral, Magistral, Ministral), specialized models such as Mistral OCR 4 for document intelligence, the Vibe agent platform for long-horizon work, Mistral Studio for building agentic systems, Mistral Forge for custom model training, and Mistral Compute for sovereign GPU cloud infrastructure. The company sells these primarily to enterprises and governments in finance, public sector, and manufacturing through API access, on-premises deployments, and cloud marketplace distribution. | Arthur Mensch Co-founder and CEO Guillaume Lample Co-founder and Chief Scientist Timothรฉe Lacroix Co-founder and CTO Jean Charles Samuelian Co-founder | API and Usage-Based Model Access Vibe Subscription Plans Enterprise Deployments and Custom Model Training |
| DeepSeek ๐จ๐ณ Hangzhou | Open-Source AI Model Development | Series A Jun 2026 | Funding $7.40B | DeepSeek develops and open-sources large language models built on deep neural networks via large-scale self-supervised pre-training and targeted optimization (SFT and RL). Its flagship DeepSeek-V4 family โ V4-Pro (1.6T total / 49B active parameters) and V4-Flash (284B total / 13B active parameters) โ uses a Mixture-of-Experts architecture with DeepSeek Sparse Attention (DSA) for 1M-token context efficiency. Models are monetized primarily through a pay-as-you-go OpenAI- and Anthropic-compatible API at platform.deepseek.com, with a free consumer web (chat.deepseek.com) and mobile app tier, and all model weights released under the MIT License. | Liang Wenfeng (ๆขๆ้) Founder and CEO Cui Tianyi Head of Harness team Zhou Guangshang Shangyan Zhou | API Token Usage (Pay-as-you-go) Free Consumer Chat (Direct-to-Consumer) Enterprise/B2B Platform Partnerships |
| StepFun ๐จ๐ณ Xuhui | Open-Source AI Model Development | Series B May 2026 | Funding $3.22B | StepFun develops and commercializes a series of multimodal foundation models (Step 1, Step 2, Step 3.5 Flash, Step 3.7 Flash, Step TTS, StepAudio 2.5 Realtime) that process and generate text, images, video, code, and speech. These models are delivered to enterprise developers through the Step Open Platform (ๅผๆพๅนณๅฐ) API and to end users through the Step AI (้ถ่ทAI) consumer assistant available across desktop, mobile, and web. | Jiang Daxin Co-founder and CEO Yin Qi Chairman of the Board | Step API Platform Consumer AI Applications Enterprise Licensing & Integration |
| Poolside ๐บ๐ธ San Francisco | AI Foundation Models for Software Engineering | Corporate Funded Aug 2026 | Funding $1.63B | Poolside builds foundation models for software engineering (the Laguna family), including the open-weight Laguna XS.2 released under Apache 2.0 alongside the proprietary Laguna M.1. The models are delivered through the Poolside Platform, which bundles an agent harness, IDE and CLI surfaces, and on-premises or VPC deployment for regulated buyers, with embedded forward-deployed research engineers. Usage-based API access is offered through platform.poolside.ai, AWS Bedrock and OpenRouter, and open weights are also distributed on Hugging Face and Ollama. | Jason Warner Co-CEO & Co-Founder Eiso Kant Co-CEO & Co-Founder Philip Drury Chief Investment Officer | Subscription Recurring Usage Based Professional Services Managed Services |
| ModelBest ๐จ๐ณ Haidian | Open-Source AI Model Development | Jul 2026 | Funding $592M | ModelBest develops and licenses the MiniCPM series of lightweight, high-performance large language and multimodal AI foundation models optimized for on-device (edge) deployment. The portfolio includes text LLMs (MiniCPM 4.0/3.0/2.0/1.0), vision-language models (MiniCPM-V 4.5/2.6/2.5/2.0), full-modal models (MiniCPM-o 2.6), and voice generation (VoxCPM). Models are delivered via open-source community distribution, enterprise licensing to OEMs and chip makers, and integration into consumer electronics, automotive intelligent cabins, embodied robots and government systems. | Liu Zhiyuan (ๅ็ฅ่ฟ) Co-founder and Chief Scientist Li Dahai (ๆๅคงๆตท) Co-founder and CEO Zeng Guoyang (ๆพๅฝๆด) Co-founder and CTO Lei Shengtao (้ทๅๆถ) COO | Model Licensing and OEM Integration Enterprise B2B Solutions |
| Stability AI ๐บ๐ธ Los Angeles | Open-Source AI Model Development | Series B Aug 2026 | Funding $412M | Stability AI develops and sells multi-modal generative AI models and enterprise platforms for image, video, audio, language, and 3D content creation. Core offerings include the open-source Stable Diffusion model family, the managed Brand Studio creative production platform for enterprise teams, a Platform API for developer integration, and a Self-Hosted License for on-premises enterprise deployment. | Emad Mostaque Founder & Chief Executive Officer Ben Brooks Head of Public Policy | Platform API Access Brand Studio Subscription Self-Hosted License |
| Sarvam AI ๐ฎ๐ณ Bengaluru | Open-Source AI Model Development | Series B Jun 2026 | Funding $349M | Sarvam AI builds a full-stack sovereign AI platform for Indian languages, developing proprietary tokenizers, open-source large language models (Sarvam-30B and Sarvam-105B), and a suite of speech, translation, and vision APIs. The company sells these capabilities through enterprise products (Samvaad voice agents, Arya enterprise AI for Work, Akshar/Vision document digitization, Studio content transformation, Indus multilingual assistant) and through self-serve developer APIs (Bulbul TTS, Saaras STT, Mayura translation, Vision document digitisation, Dubbing). It targets Indian banks, insurers, government agencies, and large enterprises needing Indic-language AI with private-cloud, on-premise, or air-gapped deployment. | Vivek Raghavan Co-founder Pratyush Kumar Co-founder | API Usage Revenue Enterprise Platform Licensing Voice Agent Platform (Sarvam Samvaad) |
| HiDreamai ๐จ๐ณ Beijing | Open-Source AI Model Development | Series C Sep 2026 | Funding $309M | HiDream.ai develops and distributes visual generative AI foundation models and multimodal AI applications. Its flagship offerings include the open-source HiDream-O1-Image model (8B parameters, MIT License), the proprietary HiDream-I1 multimodal large model (10B+ parameters), and HiDreamClaw, the first multimodal AI application purpose-built for image and video generation workflows, hosted on the vivago.ai creator studio. | API and Hosted Services Enterprise/Platform Licensing | |
| Arcee.ai ๐บ๐ธ San Francisco | Open-Source AI Model Development | Series B Sep 2026 | Funding $180M | Arcee.ai builds and releases open-weight small and large language models under the Trinity family (Nano, Mini, Large, Large-Thinking) and AFM-4.5B, all built on its proprietary AFMoE sparse Mixture-of-Experts architecture. It monetizes these models through an OpenAI-compatible hosted API (token-based pay-as-you-go consumption), an intelligent model router called Arcee Conductor, an agentic workflow platform called Arcee Orchestra, and enterprise custom-build, deployment, and integration services. Open-source toolkits (MergeKit, DistillKit, EvolKit) and Hugging Face downloads serve as a freemium funnel into the paid API and enterprise offerings. | Mark McQuade Co-founder and CEO Jacob Solawetz Co-founder and CTO Brian Benedict Co-founder and CRO Julien Simon Chief Evangelist | Token-based API consumption Enterprise model build, deployment, and integration services Model Provider redistribution |
| Nous Research ๐บ๐ธ New York | Open-Source AI Model Development | Series B Jul 2026 | Funding $145M | Nous Research develops and distributes open-source large language models (Hermes 3, Hermes 4.3), autonomous AI agents (Hermes Agent), and decentralized AI training infrastructure (Psyche Network). The company releases its core models and frameworks under MIT license for free self-hosting, while generating commercial revenue through managed cloud services (FlyHermes at $59/month), API access (Nous Portal), and enterprise offerings. | Nightwing Research Contributor Joe Li Research Contributor Phuc Nguyen Research Contributor | FlyHermes Managed Service Nous Portal API Access Open Source Distribution |
| Mindverse ๐ธ๐ฌ Singapore | Open-Source AI Model Development | Series A Jun 2026 | Funding $87.0M | Mindverse develops an open-source AI framework called Second Me that evolves with individual users' unique habits, thoughts, and preferences to create personalized AI identities, and ships a consumer-facing AI assistant product called Me.bot that represents the world's largest community of uploaded intelligences. The company focuses on building infrastructure for uploaded intelligence to protect individuality against generic 'God Model' AI systems. | ||
| Standard Intelligence ๐บ๐ธ San Francisco | Open-Source AI Model Development | Series A Apr 2026 | Funding $79.8M | Standard Intelligence develops AI foundation models, anchored by FDM-1, a foundation model for computer use trained on an 11-million-hour screen-recording video corpus that predicts key presses and mouse actions to automate any GUI-based workflow, and by hertz-dev, an 8.5B-parameter open-source autoregressive base model for full-duplex conversational speech. The company is currently pre-revenue and distributes hertz-dev via open-source channels while positioning FDM-1 for future enterprise deployment, supported by a proprietary 30-petabyte in-house storage cluster and an evaluation infrastructure capable of 1 million rollouts per hour. | Devansh Pandey Co-founder Andrej Karpathy Advisor Milan Kovac Advisor Stanley Druckenmiller Advisor | Pre-revenue / Research stage |
| Deep Cogito ๐บ๐ธ San Francisco | Open-Source AI Model Development | Series A Apr 2026 | Funding $56.0M | Deep Cogito develops and releases open-weight large language models (LLMs) trained via process supervision for reasoning chains, with its flagship Cogito v2.1 671B model positioned as the best open-weight LLM from a US company. The company distributes the model weights openly on Hugging Face, provides API inference through multiple cloud platforms (OpenRouter, Fireworks AI, Together AI, Ollama Cloud, Baseten, RunPod), supports local deployment via Ollama and Unsloth, and offers a free web-based chat interface at chat.deepcogito.com. | Drishan Arora Co-founder & CEO Dhruv Malrana Co-founder & CPO | API Access / Model Inference |
| Motif Technologies ๐ฐ๐ท Seoul | Open-Source AI Model Development | Series B May 2026 | Funding $16.0M | Motif Technologies develops and releases enterprise-focused large language models and AI foundation models. Its flagship product is Motif-2-12.7B-Reasoning, an open-weight 12.7 billion parameter reasoning LLM, and the company is also building proprietary foundation models across text, image, and video as part of South Korea's sovereign AI initiative. | Junghwan Lim Chief Technology Officer |
All 15 companies on this list build and release open-source or open-weight AI models as a core part of their offering; akta.pro tags 14 of them under the single Open-Source AI Model Development category (Poolside, a manual addition, is tagged as a Foundation Model Developer), distinguishing them from companies that merely host, fine-tune or serve third-party open models.
To qualify, a company must release its own open-weight models, not just host or serve third-party ones; candidates such as model-hosting and inference-only providers were excluded on this basis. Funding in the category is highly concentrated: the top 3 companies account for 83.2% of the $42.0B raised, and the top 5 account for 94.7%.
Company headquarters split roughly between the United States and China: 4 of 15 are based in San Francisco and 2 in Haidian (Beijing), with the rest spread one each across Paris, Hangzhou, Shanghai (Xuhui), Los Angeles, Bengaluru, Beijing, New York, Singapore and Seoul.
The list is limited to private, operating companies globally that raised a Series A or later round with a close date in 2026, tagged under akta.pro's Foundation Model Developer and Open-Source Model Ecosystem taxonomy. It excludes model-hosting and inference-only providers, and companies without their own open-weight model releases.
Companies are tagged against standard industry codes and against akta.pro's internal taxonomy, so the same set can be rebuilt by code, by segment, or by a combination of both.
NAICS 511210 Software Publishers 541511 Custom Computer Programming Services 541512 Computer Systems Design Services SIC 7372 Prepackaged Software 7371 Computer Programming Services 7373 Computer Integrated Systems Design Internal taxonomy HDAAACAA Foundation Model Developers (LLM/Multimodal Model Labs) HDAAACAM Open-Source Model Ecosystems & Model Marketplaces HDAAAC Foundation Models & Generative AI (LLMs, multimodal)Industry codes are provided for cross-referencing against public filings and procurement records. akta.pro classification is independent of them, so a company may carry an internal segment without a matching NAICS or SIC registration.