Dragon LLM

What Dragon LLM does

Dragon LLM (legal entity: LINGUA CUSTODIA SAS) is a French AI company that develops and fine-tunes generative AI language models specialized for the financial services sector and other regulated industries. Founded as a niche machine translation provider approximately a decade ago, the company has evolved into a full-stack LLM producer, rebranding to Dragon LLM in 2025 and being acquired by OVHcloud in March 2026, where it now forms the foundation of OVHcloud's newly launched AI lab. Its core technical differentiator is the proprietary "Dragon" architecture, which uses Mamba and Differential Attention mechanisms rather than traditional transformer designs, trained on EuroHPC Leonardo and JUPITER supercomputers, and explicitly engineered to be sovereign, smaller, faster, and capable of running on modest hardware with a lower energy footprint.

The company's product portfolio comprises four main lines: customized enterprise LLM solutions delivered across front office (investment research, portfolio commentary, client reporting), middle office (risk attribution, compliance monitoring, ESG analysis), and back office (document processing, regulatory filings) workflows; the VERTO document processing platform, launched September 2025 as a SaaS offering with a 14-day free trial covering data extraction, transcription, machine translation, due diligence query handling, and compliance verification; open-source models distributed via Hugging Face for community access; and the Dragon foundation architecture itself. The company monetizes through enterprise subscriptions, professional services for model fine-tuning, and a freemium open-source channel.

Dragon LLM's go-to-market is enterprise field sales targeting regulated European financial institutions, evidenced by a "Request a Meeting" call-to-action and a customer list including BNP Paribas, HSBC Private Banking, Credit Mutuel Asset Management, Intesa Sanpaolo, Natixis Investment Management, Eurizon, Kepler Cheuvreux, and CA Indosuez Wealth Management. The company holds offices in Paris (headquarters), Montigny-le-Bretonneux, and Luxembourg, has GDPR-compliant data practices under LINGUA CUSTODIA SAS, and has secured two European Commission grants (June 2024 and October 2025) totaling at least €500,000 in cash plus multi-million GPU-hour allocations for energy-efficient AI development.

Name

Dragon LLM

Legal name

LINGUA CUSTODIA SAS

Website

https://dragonllm.ai

Company type

Private

Founded year

2015

Operating status

Acquired

Headcount range

11–50 employees

Short description

Dragon LLM is a French AI company that builds and fine-tunes proprietary generative language models for financial services and other regulated industries, offering customized enterprise LLMs, the VERTO document processing platform, and open-source models distributed via Hugging Face.

Ownership category

akta.pro rank

Product category

Generative AI Software for Financial Services

NAICS

Software Publishers (5132), Custom Computer Programming Services (541511)

SIC

Services-Prepackaged Software (7372), Services-Computer Integrated Systems Design (7373)

akta.pro primary industry

Fine-Tuning, Adaptation & Custom Model Training (PEFT/LoRA/RLHF) (HDAAACAC)

akta.pro secondary industries

LLMOps & Generative AI Platforms (Prompt/Agent Orchestration, RAG) (HDAEANAD), Model Hosting, Serving & Inference Platforms (HDAAACAB), Generative AI & LLM Solutions Services (RAG, Agents, Copilots) (BPAEAHAG)

Keywords

• Specialized language models
• Financial document processing
• Sovereign AI solutions
• Generative AI finance
• Regulatory compliance AI

Headquarters

HQ city

Montigny-le-bretonneux

HQ country

France

HQ region

Europe

Offices3 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Marketing or Sales

Revenue model

1. Customized LLM Solutions: Dragon LLM provides personalized AI models tailored to specific business needs, likely through enterprise contracts with financial institutions for front office, middle office, and back office applications.
2. VERTO Platform Subscription: Document processing platform offered with a free 14-day trial, suggesting a subscription-based SaaS model for financial language processing services.
3. Fine-tuning Services: AI model fine-tuning platform services for regulated industries, enabling enterprises to customize models for specific use cases.
4. Open Source Models: Open-source models available on Hugging Face, with plans to release models once they reach sufficient performance, potentially generating revenue through support and enterprise services.

Pricing tiers

Go-to-market motion1 record

Distribution channels3 records

Marketing channels4 records

Core offering

Dragon LLM develops and sells specialized generative AI (LLM) models fine-tuned for the financial services and other regulated industries. It delivers customized AI models across front office, middle office, and back office workflows, and offers the VERTO platform for secure financial document processing including data extraction, transcription, and machine translation. Open-source versions of its models are distributed on Hugging Face.

Product overview

Dragon LLM is a full-stack LLM producer offering specialized generative AI models fine-tuned for regulated industries, particularly financial services. The product portfolio centers on two main offerings: (1) Personalised AI Models—a customized LLM solution delivered across front office, middle office, and back office tiers for investment research, compliance, and document processing; and (2) the VERTO platform—a specialized document processing platform for financial language tasks including data extraction, machine translation, and compliance verification. The company also distributes Open Source Models via HuggingFace and has developed the proprietary Dragon Architecture—an energy-efficient foundation model trained on European supercomputers that eschews traditional transformers for improved efficiency and sovereignty.

Differentiator

Problem solved

Functional benefit

Products and services

• Personalised AI Models

Quantifiable outcome

• Energy-efficient AI architecture reduces power consumption compared to traditional transformer models
• +1 more outcomes

Named customers8 records

Segments4 records

Ideal customer profiles1 record

Technology focussed Yes

API detail

Has API

No

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability5 records

Feature4 records

Partnerships

Seven partnerships are on record, tiered flagship, core and minor.

• European CommissionflagshipStrategic or Co-development PartnerDragon LLM (as Lingua Custodia) won the European Commission's Large AI Grand Challenge, receiving €250,000 and access to 2 million GPU hours on EuroHPC supercomputers. The partnership supports the development of sovereign European AI capabilities and energy-efficient language models.
• EuroHPC JUcoreStrategic or Co-development PartnerDragon LLM's models are trained on EuroHPC supercomputers Leonardo and JUPITER as part of the European High Performance Computing Joint Undertaking, providing access to world-class computational infrastructure.
• OVHcloudcoreStrategic or Co-development PartnerOVHcloud acquired Dragon LLM in March 2026 for an undisclosed amount, marking OVHcloud's first acquisition. Dragon LLM's technology forms the foundation of OVHcloud's new AI lab for sovereign generative AI services, with plans to pre-train frontier models on Jupiter supercomputer and open-source them.
• BpifrancecoreStrategic or Co-development PartnerFrench public investment bank supporting Dragon LLM as a client/partner in the French fintech ecosystem, contributing to European technological sovereignty efforts.
• France DigitaleminorGTM or Marketing PartnerEuropean tech ecosystem association representing French digital startups, Dragon LLM listed as a member/client in the ecosystem.
• France FinTechminorGTM or Marketing PartnerFrench fintech industry association supporting innovation in financial technology, Dragon LLM listed as a member in the ecosystem.
• Hugging FacecoreTechnology or IntegrationDragon LLM's open-source models are distributed through Hugging Face, providing platform access for the AI community to access and fine-tune Dragon LLM models.

Scale indicators4 records

Recent moves7 records

Expansion highlights7 records

Direct peers

• Mistral AI: Paris-based foundation model lab building sovereign European LLMs for enterprise use. Direct competitor in the same geography, targeting similar European regulated buyers with comparable sovereign-AI positioning, though with a broader (general-purpose) model portfolio rather than Dragon LLM's financial specialization.
• Aleph Alpha: German sovereign AI lab offering enterprise LLMs with emphasis on data sovereignty and regulated-industry deployment. Comparable in its European sovereignty thesis and enterprise focus, but competes from a general-purpose angle versus Dragon LLM's financial-services specialization.
• LightOn: French enterprise AI company offering foundation models and photonic-compute infrastructure for sovereign deployments. Comparable as a French sovereign AI alternative targeting regulated enterprise buyers, with overlapping go-to-market in Europe.
• Writer: Enterprise LLM platform with strong traction in financial services and regulated industries. Comparable in its fine-tuned, domain-specific LLM approach for enterprise buyers, including financial document workflows — though built on transformer architecture rather than Dragon's Mamba-based stack.
• Cohere: Enterprise-focused foundation model provider serving financial institutions and other regulated industries with secure, private deployments. Comparable in its enterprise go-to-market, focus on retrieval-augmented generation, and emphasis on data sovereignty.
• Bloomberg (BloombergGPT): Bloomberg's domain-specific LLM for financial data, news, and analytics. The most direct financial-domain LLM comparator to Dragon LLM, though built in-house for Bloomberg's proprietary terminal rather than sold externally as a general service.

Broad incumbents

• OpenAI: Frontier LLM provider whose GPT-4o/o1 models are increasingly used by financial institutions for research and document tasks. Indirect competitor via general-purpose LLMs that can substitute for Dragon LLM's specialized offerings in less regulated use cases.
• Anthropic: Frontier AI lab behind Claude, widely adopted by financial institutions for document analysis and compliance workflows. Comparable as an enterprise LLM alternative in financial services, though lacking Dragon LLM's sovereignty positioning and energy-efficiency narrative.

Emerging players

• Hebbia: AI knowledge worker platform purpose-built for financial document analysis, due diligence, and investment workflows. Comparable in its financial-document AI focus and enterprise sales motion, though positioned as an application layer atop third-party models rather than a model provider.
• Fintool: AI research copilot for institutional investors focused on financial document analysis and SEC filings. Comparable as an emerging financial-services AI application, targeting overlapping buyer personas in investment research — though smaller in scale and reach than Dragon LLM.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks6 records

Key highlights6 records

Customer concentration

• Xx.com/DragonLLM

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles3 records

Funding overview

Funding rounds8 records

Investors4 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Dragon LLM