Connect your agent to akta.pro and build company lists like this across sectors and geos.
Private open-source AI model companies, Series A or later, ranked by Q3 2026 primary-subject news volume: funding, products and services, revenue model, news signals and headcount growth for each.
| Company | Funding | Products & Services | Revenue Model | Signals (Q3 2026) | Headcount growth (Q3) |
|---|---|---|---|---|---|
| DeepSeek ↗ Foundation Model API 🇨🇳 Hangzhou, CHN | Funding $7.40B Last raised $7.40B Jun 2026 · Series A | Open-weight MoE LLMs (V4-Pro, V4-Flash) with MIT-licensed weights, served via an OpenAI/Anthropic-compatible API plus a free consumer app and embedded Tencent and Huawei integrations. | Usage Based Freemium Managed Services | 1,801 in Q3 View signals (143) → | +7% in Q3 199 → 212 employees |
| Moonshot AI ↗ Foundation Models 🇨🇳 Haidian, CHN | Funding $20.05B Last raised Undisclosed Oct 2026 · Series Unknown | Kimi family of trillion-parameter multimodal LLMs, offered as a consumer chatbot, the Kimi Work enterprise agent and an open-weight API for developers. | Usage Based Subscription Recurring Freemium | 1,330 in Q3 View signals (145) → | +4% in Q3 235 → 244 employees |
| Mistral AI ↗ Enterprise AI Platform 🇫🇷 Paris, FRA | Funding $7.50B Last raised $3.49B Sep 2026 · Series D | Proprietary and open-weight foundation models across text, code, speech and documents, delivered via the Vibe agent, Studio, Forge and a sovereign Mistral Compute cloud. | Usage Based Subscription Recurring Professional Services Managed Services Licensing Royalties | 437 in Q3 View signals (148) → | +13% in Q3 1,557 → 1,759 employees |
| Sarvam AI ↗ AI Platform 🇮🇳 Bengaluru, IND | Funding $349.0M Last raised $308.0M Jun 2026 · Series B | Sovereign Indic-language AI platform with open-source LLMs (Sarvam-30B, 105B) and voice, translation and document APIs for Indian banks, government and enterprises. | Usage Based Subscription Recurring | 166 in Q3 View signals (104) → | n/d |
| Stability AI ↗ Generative AI Platform 🇺🇸 Los Angeles, USA | Funding $412.0M Last raised $76.0M Aug 2026 · Series B | Open-weight multi-modal models (Stable Diffusion, Video, Audio, 3D, LM) licensed via API, self-hosting and the Brand Studio creative production platform for enterprises. | Subscription Recurring Usage Based Licensing Royalties | 70 in Q3 View signals (44) → | -2% in Q3 188 → 184 employees |
| Upstage ↗ Enterprise AI Platform 🇺🇸 San Jose, USA | Funding $1.09B Last raised $380.2M May 2026 · Series Unknown | Proprietary Solar LLMs paired with document-processing engines (Document Parse, Information Extract) automating underwriting, claims and KYC for insurance and finance. | Usage Based Marketplace Commission Subscription Recurring | 55 in Q3 View signals (44) → | 0% in Q3 260 → 260 employees |
| Reflection AI ↗ Foundation Models 🇮🇹 Milano, ITA | Funding $2.13B Last raised Undisclosed Apr 2026 · Series C | Open-weight frontier MoE language models and a proprietary RL training stack, with an autonomous coding system as the current flagship product for enterprises and governments. | Subscription Recurring Managed Services | 53 in Q3 View signals (31) → | +25% in Q3 268 → 336 employees |
| StepFun ↗ Foundation Models 🇨🇳 Xuhui, CHN | Funding $3.22B Last raised $2.50B May 2026 · Series Unknown | Multimodal foundation models spanning text, vision and speech, delivered via an open API platform and a consumer assistant, with automotive and robotics licensing deals. | Usage Based Freemium Licensing Royalties | 50 in Q3 View signals (25) → | n/d |
| Sakana AI ↗ AI Model Orchestration 🇯🇵 Tokyo, JPN | Funding $368.1M Last raised Undisclosed Mar 2026 · Corporate-Funded | Nature-inspired multi-agent orchestration models (Fugu, Marlin) coordinating frontier LLMs for Japanese financial institutions, defense agencies and enterprise research. | Subscription Recurring Usage Based Professional Services | 40 in Q3 View signals (33) → | +15% in Q3 244 → 280 employees |
| Prime Intellect ↗ AI Training Infrastructure 🇺🇸 San Francisco, USA | Funding $200.4M Last raised $130.0M Jul 2026 · Series A | Open-stack platform aggregating global GPU compute with 2,500+ reinforcement-learning environments, hosted post-training, evaluations and one-click inference for AI agents. | Usage Based Subscription Recurring Marketplace Commission | 40 in Q3 View signals (20) → | +16% in Q3 70 → 81 employees |
| Nous Research ↗ Open-Source AI Software 🇺🇸 Austin, USA | Funding $235.0M Last raised $90.0M Oct 2026 · Series B | Open-source Hermes LLMs, the Hermes Agent autonomous framework and the Psyche decentralized training network, distributed under MIT license with managed cloud and API tiers. | Subscription Recurring Freemium | 38 in Q3 View signals (22) → | n/d |
| Black Forest Labs ↗ Generative AI Image Models 🇩🇪 Freiburg Im Breisgau, DEU | Funding $450.2M Last raised $300.0M Dec 2025 · Series B | FLUX family of text-to-image and editing models, licensed via a commercial API, open weights and enterprise deals with Adobe, Microsoft Azure, Meta and Canva. | Usage Based Subscription Recurring Freemium | 34 in Q3 View signals (22) → | +4% in Q3 136 → 142 employees |
| Arcee.ai ↗ Open-weight Foundation Models 🇺🇸 San Francisco, USA | Funding $179.5M Last raised $150.0M Sep 2026 · Series B | Trinity family of small open-weight language models on a proprietary sparse MoE architecture, monetized via a hosted API, model router and agentic workflow platform. | Usage Based Professional Services Licensing Royalties Freemium | 26 in Q3 View signals (12) → | +7% in Q3 42 → 45 employees |
| Deep Cogito ↗ Foundation Models 🇺🇸 San Francisco, USA | Funding $56.0M Last raised $43.0M Apr 2026 · Series A | Cogito, an open-weight reasoning LLM trained via process supervision, distributed on Hugging Face and through multiple inference clouds plus a free chat interface. | Usage Based | 23 in Q3 View signals (7) → | +38% in Q3 13 → 18 employees |
| Motif Technologies ↗ Enterprise AI Foundation Models 🇰🇷 Seoul, KOR | Funding $16.0M Last raised $16.0M May 2026 · Series B | Motif-2-12.7B-Reasoning, an open-weight reasoning LLM, built as part of South Korea's sovereign AI initiative for government and industrial enterprise customers. | n/d | 23 in Q3 View signals (21) → | n/d |
An open-source AI model company trains foundation models and releases the weights publicly, so anyone can download, run and fine-tune them. The category is also called open-weight model labs or open AI labs.
Frontier open-weight labs such as DeepSeek, Moonshot AI, Mistral AI, StepFun and Reflection AI compete with the largest closed models on reasoning and coding.
Sovereign and regional labs such as Sarvam AI, Upstage, Sakana AI and Motif Technologies build models for local languages and markets. Specialist labs such as Black Forest Labs and Stability AI focus on image and media generation, while Nous Research, Arcee.ai, Deep Cogito and Prime Intellect build smaller or community-trained models.
The category is defined against closed-model labs that only sell API access. Open-weight companies give away the model and earn money from hosted APIs, enterprise deployments, licensing and services.
Revenue models in the table reflect that split: 12 of 15 companies price at least partly on usage through hosted APIs, and many add licensing or managed deployments.
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.
NAICSIndustry 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.