Unlimited Basic grows. Your price stays at €14.95.

7 November update: we have reduced credit consumption for DeepSeek and GLM 5.3. Explore how many tokens Unlimited Basic now covers and the lower API prices.

A productive session with AI rarely ends with the first answer. You ask for an approach, try a change, find an error and have another go. You add context. Explore another solution. That is how an idea starts becoming something you can use.

When every iteration adds tokens to the bill, it is easy to start watching consumption in the middle of your work. If you also need to transcribe a recording, search your documents or switch models, there are more tools to configure and more decisions to make.

We want you to have room to work. That is why Unlimited Basic is growing while keeping its €14.95 monthly price. It now includes two unlimited Qwen models, subject to fair use, and nine premium models sharing your 5,000 monthly AI Credits. You can also use it through QDivZero Harness, our free application for working with models, files and agents.

Users already working with Unlimited Basic love the plan. That response encourages us to keep expanding what they can do for the same monthly price. This update brings more options to the tasks that make up their working day.

Two unlimited Qwen models to keep your project moving

Some tasks take several attempts. A function still needs work, a document deserves another revision or an agent needs more context before continuing. Having a model you can return to throughout that process changes how you work.

Qwen 3.8 27B remains included, and we improved its serving performance this week. Alongside it, we are adding Qwen 3.6 35B-A3B with unlimited use, subject to fair use, so you can choose between them for each task.

Qwen 3.6 uses a mixture-of-experts architecture with 35 billion total parameters and around 3 billion active parameters. Its official model card highlights coding and agent work among its capabilities. You can read its evaluations and specifications in the Qwen3.6-35B-A3B release.

Try each model on a code review, an explanation or a multi-step task and choose the one that suits your workflow. Neither uses AI Credits, and both run in Europe. Unlimited access is subject to fair use, availability and plan limits, which you can review in the Unlimited Basic conditions.

Nine premium models for what comes next

One project can need very different capabilities. After coding, you might need to turn an interview into text, find a document related to a query or explore another model’s approach to a problem.

Your 5,000 monthly AI Credits let you bring those capabilities into the same account. The plan includes five chat models and four other options for embeddings, transcription and decisions.

Build a workflow around the task you actually need to complete. Transcribe a meeting with Whisper, work on that text with one of the Qwen models and add semantic search when you want to retrieve related documentation. Qwen3 Embedding 8B supports retrieval based on the meaning of text and code.

You can also reserve some credits for another chat model when you want to explore an alternative. Having several options available makes it easier to assess what adds value to your application using its own data and tasks.

To give you a sense of scale, allocating the entire monthly allowance to GLM 5.3 Flash could cover around 999 ten-turn conversations under our estimate assumptions. For transcription, the reference is around 34 hours with Whisper Large V3 or 108 hours with Whisper Large V3 Turbo.

These are separate scenarios, each using the full 5,000 AI Credits. The estimates are not additive. All premium models share one allowance, and actual usage depends on the task, context and configuration. The Unlimited Basic models and usage section explains the token assumptions and includes examples for coding, books and decisions.

Credits renew with your monthly billing cycle. If you use them up, both Qwen models remain available under their unlimited-use conditions.

QDivZero Harness brings it into your working environment

The next step often comes when you want AI to work on your project. It needs access to files, an understanding of what you are doing and a way to propose changes you can review.

QDivZero Harness brings that workspace into a native Mac application. Open a folder, connect a model and describe what you want to achieve. The agent can inspect and edit files, use the terminal and research the web. Follow its activity in the conversation and review context and changes through the inspector.

Imagine you are building a small application. You can ask for help understanding the project, work through an implementation and review changes before incorporating them. If you are researching or preparing a proposal, generated documents can also open inside the conversation. AI becomes part of the environment where you are working.

Harness also includes skills, per-project MCP connections, memory and delegated agents. These help you adapt the workspace to your tools and continue tasks that span more than one conversation. The capabilities of this version are described in the QDivZero Harness project.

The application is free and open source under the MIT license. Connect it to Unlimited Basic to use the plan’s models. You can also connect other providers or local models through Ollama without a QDivZero subscription. Connected services retain their own usage terms.

The installer is available for Apple Silicon Macs running macOS 13 or later. Download Harness and follow the installation steps to set up your first connection.

More options for the same €14.95 a month

The upgrade keeps the €14.95 monthly price, VAT included, with no minimum commitment. That amount covers the access defined by Unlimited Basic, including both unlimited Qwen models subject to fair use and 5,000 shared AI Credits for the nine premium models.

You can activate the plan without topping up a balance or deploying a GPU. Then use the model identifiers shown in the platform’s Unlimited section to connect the included models to your applications or Harness.

Subscription and balance can coexist in the same account. Use the plan for your regular work and pay for other services when your project needs them. Public models accessed through balance and dedicated deployments have their own rates.

DeepSeek prices fall by 50% as the public catalog grows

The update also reaches people integrating AI into applications who prefer to pay according to usage. We have halved the rates for all our public DeepSeek models.

We are also expanding the public catalog with CLEF Flash, Whisper, Qwen3 Embedding 8B, GLM 5.3 and DeepSeek V4.1 Flash 1017. Here are some of the announced prices.

An hour of audio means the duration of the recording being processed. An execution second measures compute time, while tokens measure the content processed by the model. Each rate uses the unit appropriate to its service.

We are continuing to optimize inference infrastructure too. The announced compute references are €6 per hour for GLM 5.3, from €4 per hour for GLM 5.3 Flash and from €4 per hour for DeepSeek V4.1 Flash.

These public-access and compute prices are separate from the Unlimited Basic subscription. You can check a deployment’s cost before starting it. Billing continues while it remains active, even without requests. Review public model prices and how deployments work to choose the approach that suits your workload.

Start with something you already want to finish

Think of a code change you want to complete, an automation you have been putting off or a recording you need to turn into something useful. You already have a starting point for trying the update.

Open QDivZero and go to Unlimited to view or activate the plan. To connect an application, create your API key and follow the integration guide. For desktop work, set up your connection in Harness and choose an included model.

We are preparing new subscriptions and products on the platform while continuing to publish our open-source work through Valendra Tech. In the meantime, this upgrade gives you more room to start, experiment and make progress for the same monthly price.

Frequently asked questions about the Unlimited Basic upgrade

Which models have unlimited use?

Qwen 3.8 27B and Qwen 3.6 35B-A3B. Both run in Europe and use no AI Credits. Access is subject to fair use, availability and plan limits.

Does each premium model get 5,000 AI Credits?

The nine premium models share a single allowance of 5,000 monthly AI Credits. Allocate them according to your needs. When they run out, wait until the next monthly renewal to use premium access through the plan again.

Is the subscription price increasing?

Unlimited Basic remains €14.95 per month, VAT included, with no minimum commitment. Monetary balance for other public models or dedicated deployments is managed separately.

Do I need Unlimited Basic to use Harness?

Harness is free and independent of the subscription. Connect it to the plan, other compatible providers or local models through Ollama. Each connection’s usage depends on the service you use.

Explore Unlimited Basic and its included models. Choose what you want to work with and start shaping your next project.