Monthly subscription
Available plans
Start with Unlimited Basic and access the included models through the API without setting up your own deployment.
- Usage and limits defined in the plan.
- No servers or GPUs to manage.
- No subscription commitment.
QDivZero EnterpriseYour AI project,with room to grow.
For startups building their product and businesses bringing AI into their operations. Start with an available plan, use models billed by tokens or deploy on dedicated compute. If you need something tailored, we will shape an Enterprise plan with you.
- The entire Hugging Face catalogue
- Private and fine-tuned models
- Unlimited users, no additional charges per user
For businesses and startups
Start with what you need today.
Go straight to QDivZero and choose how to use the platform. Combine a subscription with a balance for public models and dedicated deployments.
Pay by tokens
Ready-to-use models
Add funds and access public models through an OpenAI-compatible API. Usage is deducted from your balance at each model’s rates.
- Billed for input and output tokens.
- No subscription required for this option.
- One balance across different models.
Pay by compute
Your dedicated deployment
Deploy a compatible Hugging Face, private or fine-tuned model on dedicated capacity. Check the price before starting your deployment.
- Compute prorated per active second.
- Billed while active, even without requests.
- Region and GPUs depend on requirements and availability.
QDivZero Enterprise
Discuss my tailored planA plan shaped around your business.
If your project needs different capacity, a specific configuration or tailored commercial terms, talk to us. We review your usage volume, models and deployment requirements to prepare a QDivZero proposal for your business or startup.
- Capacity and usage volume
- Configuration and regions
- Terms agreed with you
We agree on configuration, terms and pricing before starting. Any changes are quoted and approved by you before being applied. IT consulting, development, maintenance and 24/7 coverage are agreed separately.
Our IT consulting clients already work with QDivZero.
We develop the platform at Valendra and support businesses that need a plan adapted to their project.
Freedom to choose models
The right model exists. It doesn’t have to be on a shortlist.
The entire Hugging Face catalogue is your starting point, alongside your private and fine-tuned models. From a small classifier to a multimodal application: choose what your business needs.
All of Hugging Face
Explore open-weight models and their variants to find the ones that fit your workload.
Your models and fine-tunings
Bring private or specialist weights to a compatible dedicated inference environment.
Models already available by API
Start with public models or those included in Unlimited Basic without creating your own deployment.
Text and codeImages and videoSpeech and audioVisionEmbeddingsMultimodalBiomolecular modelling
For each deployment, QDivZero validates architecture, runtime and memory, then finds compatible capacity across providers. Model licensing and availability in your selected region also determine what you can run.
Explore model deploymentA common foundation for your teams
From a model to an application that works for your business.
Bring each AI project’s building blocks together and avoid rebuilding infrastructure whenever a model, provider or use case changes.
Compute that fits the model
QDivZero identifies requirements, finds compatible GPUs across providers, prepares the runtime and delivers an endpoint. Choose from available regions, including Europe.
Deployments · GPUs · RegionsOne API for your applications
Connect OpenAI-compatible clients and SDKs. Evaluate alternatives and change the model behind the endpoint, checking the capabilities each application uses.
OpenAI-compatible APIYour data becomes context
Combine embeddings, semantic search and Retrieval to build RAG over documents, contracts or company knowledge. Decide what to store and how long to retain it.
Retrieval · Embeddings · RAGEach request, the right model
Use LLM Model Routing to define priorities and fallbacks based on cost, latency, quality or availability. Maintain a common layer for distributing requests.
LLM Model RoutingPolicies around inference
LLM Firewall lets you apply guardrails against prompt injection, jailbreaks, prohibited content or sensitive data according to your application’s policies.
LLM Firewall · GuardrailsMore teams, one platform
Bring development, data, product and operations together with unlimited users and no additional charges per user. Share a foundation of models, endpoints and tools.
Unlimited usersQDivZero, by Valendra
See what your team puts into production.
QDivZeroReady to work with AI: manage your models, deployments and usage from QDivZero. An OpenAI-compatible API also lets you use the models from your applications and tools.
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.
Private AI, with explicit choices
You choose the model. You also define how data is handled.
Residency for each workload
QDivZero is developed and operated by Valendra, a Spanish company. Unlimited runs in Europe; for your own deployments, choose an available region. Check each service’s location if you require European residency.
Zero logs and Responses context
Inference content is not logged or used for training. Responses is the exception: it retains context to continue conversations. You can delete it or disable logs globally; in that case, use Chat Completions.
Encryption in transit
HTTPS/TLS encrypts communications between your application, QDivZero and the compute or inference provider. It protects data in transit in both directions. Your prompts and responses are not sent to the creators of the models.
From platform to outcomes
Start with a specific business problem.
Every new user adds to your AI bill.
Compare subscriptions, token billing and dedicated compute against your product’s usage. Routing lets you reserve more powerful models for the tasks that need them and evaluate the cost of each task.
Your team spends more time on GPUs than on the product.
Deploying a model means finding capacity, preparing the runtime and maintaining an endpoint. QDivZero manages that deployment so your team can focus on the features your customers are waiting for.
You want another model, but switching means rebuilding integrations.
Evaluate alternatives through an OpenAI-compatible API and reuse your clients and SDKs. Switch or combine models with fewer application changes, checking the capabilities each workflow uses.
Every team signs up for a different AI service.
Product, engineering and operations end up with separate accounts, providers and integrations. Bring models, endpoints and usage into one platform, with unlimited users and no additional charges per user.
Customers ask where their data is processed.
Choose each deployment’s region from those available, including European regions. QDivZero supports private and fine-tuned models on dedicated compute, with LLM Firewall to apply protections to requests.
A model stops responding and your service suffers.
Configure priorities and fallback models with LLM Model Routing so you have an alternative when a destination is unavailable. Distribute requests according to the capabilities, latency or cost each application needs.
Frequently asked questions about QDivZero Enterprise
Is QDivZero for businesses and startups?
Yes. Startups can use QDivZero to develop and launch their product, while businesses can use it in their applications and workflows. Start directly with a subscription, public models billed by tokens or dedicated deployments billed by compute. If you need specific capacity, configuration or terms, contact Valendra for a tailored Enterprise plan.
How can I start without requesting a quote?
Go to QDivZero and choose the option you need: a subscription for the models and usage included in the plan, a balance for public models billed by input and output tokens, or a dedicated deployment billed by active compute time. You can combine these options. Dedicated compute draws from your balance while the deployment is active, even without requests.
Does QDivZero Enterprise include the entire Hugging Face catalogue?
You can explore all of Hugging Face and choose models to deploy, as well as bring your own private or fine-tuned weights. Ready-to-use API models are another option, not the product’s limit. Each deployment depends on architecture and runtime compatibility, memory, licensing and available capacity. Compute is billed for the selected configuration; it is not automatically included in Unlimited Basic.
What infrastructure does QDivZero manage when deploying a model?
The platform identifies model requirements, validates memory and runtime, finds compatible capacity across providers and prepares the inference environment. It then exposes an endpoint for your application. You can choose from available regions and monitor resource usage and performance from the platform.
Can I build enterprise RAG with private documents?
Yes. Combine embedding models, Retrieval and LLMs to index documents and retrieve relevant context during inference. Data you choose to store for RAG requires defined permissions, location and retention. Valendra can prepare the integration, evaluate answers and connect it to your company’s information sources.
How does QDivZero protect prompts and responses?
QDivZero applies zero logs to inference content, does not use it for training and does not send it to model creators. The Responses API retains context for continuing conversations: delete it or disable logs globally and use Chat Completions. Communications use HTTPS/TLS. Final privacy also depends on the region and how your application uses and retains data.
Is the QDivZero API compatible with OpenAI?
Yes. Reuse OpenAI-compatible clients and SDKs by adjusting the endpoint, credentials and model identifier. Validate the specific capabilities your application uses, such as tools, modalities or output formats, when switching between models.
Is QDivZero Enterprise billed per user?
The platform supports unlimited users in one account without additional charges per user. Service costs depend on the mode: subscription, tokens for public models or active compute time for dedicated deployments. Adding users does not make model or infrastructure usage unlimited.
When should I contact you for a tailored Enterprise plan?
When the available options do not fit your company’s volume, capacity, configuration or commercial requirements. We review models, regions and deployment needs with you to prepare a proposal. You do not need to contact us to use published plans, token-based models or the dedicated compute available in QDivZero. Additional IT consulting, development and maintenance work is agreed separately.
Your next step with AI starts here.
Explore QDivZero plans or tell us what your business needs. If you are looking for a tailored Enterprise plan, we will help you define it.