01
The right model for your process
Evaluate models for assistants, document extraction, classification or semantic search. Models already available through the API are just one way to get started.
Private AI in Europe. Your models. Your data. Your control.
Bring AI to your documents, applications and processes with QDivZero, Valendra’s platform. Explore the entire Hugging Face catalogue or use your own models, run them in European regions and connect your systems without managing GPUs.
- Hugging Face and your own models
- Your choice of European region
- OpenAI-compatible API
From problem to solution
From an AI experiment to a solution that fits your business.
The model is one part. Your data, processing location, integration and operation matter too.
What is holding you backWhat we resolve
What is holding you backThe model you need is missing from your provider’s small catalogue.
What we resolveExplore all of Hugging Face and deploy compatible, private or fine-tuned models with QDivZero.
What is holding you backYour documents and requests must be processed in Europe.
What we resolveWe define the services, European deployment region and data retention rules for your workload.
What is holding you backYour team needs to solve a process as well as access a model.
What we resolveWe connect the API to your systems, add RAG or agents where useful and validate the results.
Freedom to choose models
The entire Hugging Face catalogue. Your business sets the criteria.
Explore the entire Hugging Face catalogue: text, code, images, video, audio, vision, embeddings and specialist models. QDivZero checks requirements and finds compatible compute. You can also deploy private and fine-tuned models, subject to licensing, architecture and available resources.
Hugging FaceYour own modelsMultimodal
02
Inference in the right region
Select a European region for deployments that require European residency. The platform prepares the runtime, finds compatible GPU capacity and provides the endpoint.
03
Privacy built into the integration
Your content is not used for training or sent to model creators. Inference uses zero logs; if you use Responses, configure context retention to match your requirements.
QDivZero
Ready-to-use API models or your own private deployment.
01
Start with models already running
Connect your application through an OpenAI-compatible API. Use Unlimited Basic or public models billed by usage; check each service’s region against your European residency requirements.
For getting started without preparing your own deployment.
Explore QDivZero Enterprise02
Deploy from Hugging Face or with your weights
Choose from the entire Hugging Face catalogue, private or fine-tuned models. QDivZero validates runtime and memory, finds compute across providers and prepares an endpoint in your chosen available region.
For specialist models, dedicated capacity and specific requirements.
Design my tailored planWhat changes for your business
Useful AI for your business. Clear decisions about your data.
Europe, by choice
Unlimited runs in Europe. For other services, we select and verify the location your project requires.
Your content stays out of training
Your prompts and responses do not train models. Inference uses zero logs, with a configurable exception for Responses context.
Connected company knowledge
Combine embeddings and Retrieval so an assistant can retrieve context from your documents with RAG.
One integration, more options
An OpenAI-compatible API, routing across models and LLM Firewall to apply your application’s policies.
Data residency depends on the service and region you choose.
QDivZero, by Valendra
We build QDivZero. We understand what runs behind your inference.
QDivZeroWe develop the platform connecting models, compute and AI tools. Use QDivZero directly with your team or bring us in for a specific need: evaluating models, configuring a European deployment, preparing RAG or integrating an endpoint into your application. We can also design and build a solution with an agreed scope.
Deployments
Select a model to explore its monitoring.
6 active deploymentsOpen the platform
MonitoringQwen3.8 27B
Qwen3.8 27B
121.4tokens / s
0 errors186 ms P95
GPU
64%VRAM
42 / 96 GBRequests · 24 h
18,420RegionEuropeGPURTX PRO 6000Configurable privacy
Select another model to explore its metrics.
Plans and usage options
Choose how to start with private AI.
Start with QDivZero’s published options or tell us what you need. If your project requires different capacity, integration or configuration, Valendra can design and prepare a tailored plan.
Unlimited users for your team.One account for your whole organisation, with no additional charges per user.
Monthly subscription
Unlimited Basic
€14.95/ month
VAT included · No minimum commitment
Integrate the included models through an API, without deploying GPUs.
- Unlimited Qwen 3.8 27B, subject to fair use.
- Qwen: 60 requests/min and 5 concurrent requests.
- 5,000 monthly AI Credits shared across the GLM and DeepSeek models in the plan.
- The plan’s models run in Europe.
- OpenAI-compatible API and ticket-based support.
Pay with balance
Public models
Per token
Based on model and usage
Access other ready-to-use models and pay for the requests you make.
- Add euro balance to your account.
- Billing for input and output tokens.
- No subscription required for this option.
- Check the model’s region if you need EU data residency.
Balance for compute
Dedicated deployment
For compute
Based on model, resources and region
Deploy compatible models from the entire Hugging Face catalogue, your own weights or fine-tuned models on dedicated capacity.
- See the price before deploying.
- Compute prorated per second of activity.
- Active deployments use balance even without requests.
- Region and capacity subject to availability.
Need an AI configuration tailored to your project?
Ask us to validate a model, prepare RAG, integrate an endpoint or deploy a workload. If you need a broader solution, we can design it together. We agree the scope, what your team contributes and what we deliver.
- Capacity and concurrency
- Models and integration
- Region and infrastructure
Your questions, answered
Frequently asked questions about private AI in Europe and QDivZero
Costs, integration, data and operation: practical answers before your next step.
Can I deploy any model from the Hugging Face catalogue with QDivZero?
You can explore the entire Hugging Face catalogue, as well as use your own private and fine-tuned models. Running a specific model depends on its architecture, runtime, memory, licence and available capacity in the selected region. QDivZero checks requirements and finds compatible compute; you are not restricted to models already served through its API.
What does private AI mean for businesses operating in Europe?
It means defining where models run, who processes information and how content is retained. QDivZero lets you choose European regions for private deployments, use an OpenAI-compatible API and keep your data out of model training. Valendra integrates that foundation into your systems and reviews your access, retention and residency requirements.
Can I use DeepSeek or Qwen without sending data to China?
Yes, by running their open-weight versions on compatible European infrastructure. Downloading and serving their weights does not require using their creators’ APIs. QDivZero does not send your requests to the companies that developed those models; choose a service or deployment region that meets your project’s requirements.
What does zero logs mean and how does the Responses API work?
QDivZero does not log inference content or use it for model training. The exception is the Responses API, which retains context to continue conversations. You can delete that context or disable logs globally; with global logging disabled, Responses is unavailable and you must use Chat Completions. Documents you choose to store in Retrieval have their own retention lifecycle.
How can I connect internal documents to private AI with RAG?
We prepare documents, generate embeddings and configure Retrieval to find relevant passages before querying the model. This supports assistants over contracts, procedures or internal documentation. We agree permissions, location and retention for that content and evaluate answers using real examples from your business.
Can I run a private or fine-tuned model on dedicated infrastructure?
Yes. QDivZero supports deploying your own private and fine-tuned weights on compatible dedicated capacity. Check licensing, memory and runtime requirements and available regions. The platform prepares the environment and exposes an endpoint for your application; Valendra can help evaluate and integrate the model.
Do all QDivZero services process data in Europe?
Unlimited runs in Europe. Public model regions depend on the service, while your own deployments let you select from available locations. For this solution we confirm a suitable European region before processing information. A European platform does not replace checking each workload’s residency or your application’s GDPR obligations.
Do I need to rebuild my application to use a private AI API?
The OpenAI-compatible API allows you to reuse familiar clients and SDKs, adjusting credentials, endpoint and model. We check the features, tools and response formats your application uses for compatibility, then evaluate quality, latency and cost with the tasks your application actually performs.
What plans and pricing does QDivZero offer for AI API access?
QDivZero offers Unlimited Basic at €14.95/month, VAT included and without a minimum term: Qwen 3.8 27B unlimited subject to fair use and 5,000 monthly AI Credits shared by the included premium models. Public models use token-based balance billing and dedicated deployments use active compute time. Check current models, rates and terms on qdiv0.com.
Can I combine Unlimited Basic with balance for other AI models?
Yes. Subscription and balance can coexist in one account. Included Unlimited Basic usage does not deduct euro balance. Premium AI Credits renew monthly and are separate from balance used for other public models and deployments at their applicable rates. The subscription does not include deploying the entire Hugging Face catalogue.
Can you design and prepare a tailored private AI plan for my business?
Yes. We review models, documents, traffic, concurrency, residency and integrations to prepare a tailored proposal. Scope and price are agreed before work begins. Platform usage, integration and subsequent services are identified separately; solution maintenance and 24/7 coverage are quoted and contracted separately.
Can you help with just model selection, RAG or private AI integration?
Yes. You can use QDivZero directly and engage us for one area: model evaluation, Retrieval and RAG preparation, deployment in a European region or API integration. We review what your team already has covered and agree our intervention, deliverables and price. We can also take on a broader scope if needed.
Is there a user limit or a per-person fee for my team?
QDivZero supports unlimited users in one account, with no additional charges per user. Bring your whole team onto the platform without buying a separate licence for each person. Usage follows the selected mode: fair use and AI Credits for the subscription, tokens for public models or active compute time for dedicated deployments.
Your AI, with a plan that fits.
Use QDivZero with your team or tell us which part needs work: models, data, integration or deployment. We define the scope of our collaboration together.