PRICING
See how QDivZero works, explore pay-as-you-go rates or choose a subscription. A European platform that does not send your data to model providers or store your prompts and responses.
SUBSCRIPTION
€14.95/month
VAT included · No minimum commitment
For connecting models to your applications without deploying infrastructure.
BALANCE
Balance in euros · Usage-based pricing
Unlimited runs in Europe · Zero logs across every access option · Technical support via tickets
HOW PRICING WORKS
The subscription covers the models and usage included in the plan. Balance in euros lets you use other models or run your own deployments. Keep both in the same account, with no additional per-user cost and no balance deductions for usage included in the subscription.
HOW BILLING WORKS
Balance in euros and AI Credits are independent. Add balance to pay for usage. AI Credits come with the subscription and replenish with each monthly renewal, without converting into balance in euros.
Included in all our plans at no extra cost. You only pay for the models used to evaluate, route and respond, according to their billing option.
With Unlimited, you can use Qwen 3.8 27B or included premium models under their respective usage and AI Credit terms. Models used through balance are billed at their published rates.
Retrieval storage is billed at €0.001 per GB per hour, based on the space used and storage duration.
Embedding generation is billed separately at the rate of the model you choose.
Answers about balance, subscriptions, public models, and open-weight deployments.
Create a free account and explore the models. Activate a subscription for included usage or add balance for other models and deployments. Combine both options and contact technical support through tickets if you encounter an error.
QDIVZERO BALANCE
Add balance to your account and decide where to use it. The same balance gives you access to ready-to-use public models and dedicated compute for open-weight models. You do not need to change plans to change models.
Pay per token
Use models that are ready through an OpenAI-compatible API. Input and output tokens are deducted from your balance according to each model's published rate.
Pay for compute
Choose an open-weight model from Hugging Face and deploy dedicated infrastructure. Compute is deducted from the same balance while the deployment remains active.
Privacy in every region, for both public models and dedicated deployments. We apply zero logs: we do not log your prompts or responses. We do not use them to train models or send your data to model creators. Communications are encrypted over HTTPS/TLS between your application and QDivZero, and between QDivZero and the GPU or public inference provider, in both directions.
Explore modelsPAY FOR COMPUTE, NOT TOKENS.
Choose an open-weight model from Hugging Face and configure your deployment requirements. QDivZero checks the infrastructure it needs and searches for compatible capacity across providers, prioritizing the best available price-to-compute ratio. The cost is deducted from the same account balance.
With a dedicated deployment, you pay for compute while it is active, not for each input or output token.
View providersSearch for a Hugging Face model. The calculator estimates the VRAM it needs and checks compatible infrastructure at current GPU-second prices.
Search for a model or select a quantized GGUF variant to estimate VRAM and deployment cost.
COMPUTE PRICING
A GPU price can vary by provider, region, and availability. QDivZero therefore does not rely on a single infrastructure source. We find capacity compatible with your model and requirements and prioritize the best available price-to-compute ratio.
For reference, Replicate currently lists L40S at $0.000975/s, A100 80 GB at $0.001400/s, and H100 at $0.001525/s. QDivZero searches across providers before proposing where to run your model.
Pay per token from €0.07/M.
You do not need a different integration for every provider. Access DeepSeek V4.1 Flash, Qwen 3.8 27B and other models in the table through a single OpenAI-compatible API, with one balance and no infrastructure to deploy. Try one model, switch to another, or use several in your applications without integrating each provider from scratch.
All public models are deployed in Europe, except GPT-6 models, which are deployed in the United States.
Explore modelsUsage is deducted automatically from your account balance.
Prices updated 10/23/2026, 08:55:45 UTC
READY-TO-USE MODELS · RUN IN EUROPE
Connect Qwen 3.8 27B, GLM-5.3-Flash, DeepSeek V4.1 Flash and DeepSeek V4 Flash 0731 through an OpenAI-compatible API. Unlimited models run in Europe, with zero logs of inference content and HTTPS/TLS encryption. No GPU management. €14.95/month, VAT included, with no minimum commitment.
Qwen 3.8 27B is unlimited, subject to fair use. The three premium models share 5,000 monthly AI Credits. If you use them up, wait until the next monthly renewal to use premium models with the plan again. Qwen 3.8 27B remains available. Balance for other services is managed separately.
API access for your applications and compatible tools. This is not a chat application. First create an account, then subscribe to Unlimited Basic and connect your API key.
Create your free account and subscribe when you are ready. No balance top-up is needed for included usage.
Get your key in the platform and the ID of the included model you want to use.
Set https://api.qdiv0.com/v1 as the base URL, add your key and select the model. Send your first request.
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["QDIV0_API_KEY"],
base_url="https://api.qdiv0.com/v1",
)
response = client.chat.completions.create(
model="unlimited:qwen/qwen-3.8-27b",
messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)Install the openai package and set QDIV0_API_KEY to your key. Included usage requires an active subscription. Running this example sends a real request.
What the subscription includes
€14.95/month
Yes: €14.95/month, VAT included, with no minimum commitment. Cancel your subscription in the platform to stop next month’s charge. Cancellation does not refund the period already paid.
What you can build
Qwen 3.8 27B is unlimited, subject to fair use. It uses no AI Credits.
You receive 5,000 AI Credits per month for GLM-5.3-Flash, DeepSeek V4.1 Flash and DeepSeek V4 Flash 0731.
All premium models share one allowance. Using more credits on one leaves fewer for the others.
AI Credits are usage credits included in your subscription. They are not money or a QDivZero balance top-up. Monetary balance is used separately for other models and services at their published rates.
The latest frontier models are still billed per million tokens. That works for occasional use, but costs grow with conversations, agents, long contexts, and automations.
The more you use it, the more sense it makes to stop counting tokens.
Prices checked on September 30, 2026. *Gemini 3.8 Flash keeps this introductory price through December 31, 2026. Included Qwen 3.8 27B usage has no per-token billing and is subject to fair use: 60 requests per minute per account/API key and 5 concurrent requests. The compared models have different capabilities and features.
In Qwen's published evaluation, Qwen 3.8 27B is compared with Claude Opus 4.6 Max as a frontier reference. It scores higher on SWE-bench Pro, CoWorkBench, and LiveCodeBench v6, while Opus retains a slight advantage on GPQA Diamond. The results position Qwen 3.8 27B as a competitive option for coding and agentic workflows without implying overall equivalence between the two models.
Qwen publishes these results on the model's official page. They are not from an independent evaluation and may vary depending on each benchmark's setup and testing conditions.
Official source: Qwen3.8-27B model cardThree premium models to complement Qwen 3.8 27B, included in Unlimited Basic’s shared allowance of 5,000 AI Credits per month.
Code, vision and tools
Combines coding, image understanding and tool use. Useful for assistants that need to interpret information and act on it.
Text, images and context
Works with text and images in the same task. An option for analysing documentation and reasoning with context across multiple steps.
Coding agents
The 0731 release strengthens agent and tool use. Built for multi-step coding tasks and work across repositories.
Unlimited runs in Europe, while deployments let you choose from the available regions. In every case, we apply a zero logs policy: we do not log your prompts or responses. We do not use them to train models or send your data to model creators, regardless of the region selected.
How we protect your dataCancel your subscription in the platform to stop next month’s charge. Cancellation does not refund the period already paid. If AI Credits run out, wait for the monthly renewal. Qwen remains available under fair use.
See how the plan worksIf you encounter an error, open a ticket in our support portal. Our technical team will handle the issue according to the ticket’s priority.
Open a support ticket