Skip to content

The Dutch inference provider for open models

Give your customers access to open models running on infrastructure in the Netherlands.

Get API accessSee all pricing

Pick a model and send a request

Input / output, per one million tokens

Streamed by DeepSeek V4.1 Flash, an open model running on infrastructure in the Netherlands.

The models

Our village of models. On Dutch infrastructure.

Bring open-weight models from leading AI labs to your customers, powered by infrastructure in the Netherlands.

  • DeepSeek
  • Qwen
  • Meta
  • Mistral
  • OpenAI
  • z.ai
  • Moonshot
  • MiniMax

Why OxVillage

Customer requests in.
European inference out.

OpenAI-Compatible API

Point any OpenAI SDK at our /v1 endpoints. Chat completions, embeddings, streaming, tool calling — zero migration code.

A Dutch destination for your router

Add OxVillage as an inference provider and route customer requests to open models running in the Netherlands.

Optimized Serving

vLLM and SGLang with continuous batching, quantization and per-model tuning. More tokens per GPU-hour — a lower cost per million.

Wholesale Economics

Per-token pricing built for routers and platforms. Reserved capacity and bare GPU available as you scale.

Integration

One line to switch.
Zero lock-in.

For Routers & Aggregators

Your customers choose a model; your platform chooses where each request runs. Add OxVillage as a Netherlands-hosted inference provider for open models. Route traffic to us as a primary provider, a fallback, or when a customer needs European processing. You keep control over model selection, routing and billing while giving your customers another reliable destination.

POST /v1/chat/completions
GET /v1/models
GET /v1/embeddings
Explore provider integration

Point your SDK at us.

The OpenAI SDK you already use, executed in the Netherlands. Change the base URL to api.oxvillage.nl — keep your code, prompts and tooling exactly as they are.

from openai import OpenAIclient = OpenAI(    api_key="YOUR_API_KEY",    base_url="https://api.oxvillage.nl/v1",)response = client.chat.completions.create(    model="YOUR_MODEL",    messages=[{"role": "user", "content": "Hello"}],)

For AI Platforms

Build chatbots, copilots and other AI features into your product without operating model infrastructure yourself. Use OxVillage's OpenAI-compatible API to access open models running in the Netherlands. You stay in control of the experience your users see and can tell customers where their AI requests are processed.

European inference for your product

For Developers

Build with open models while keeping inference on infrastructure in the Netherlands. Point your existing OpenAI SDK at OxVillage's /v1 endpoint to work with sensitive prompts, documents and customer data without deploying GPUs yourself. Choose the models your application needs and keep control over where model execution happens.

Built in Europe · OpenAI-compatible API

Connect in three steps.

Supported features and limits depend on the selected model and your access configuration.

Discuss your integration

  1. Request access

    Tell us which models, traffic volumes and processing requirements you need.

  2. Configure your client

    Install the Python client with pip install openai, then use the API key, base URL and exact model identifier supplied with your access.

  3. Test your workload

    Check responses, latency and model-specific features before routing production traffic. Keep API keys on your server.

Pricing

Pay per token.

Transparent model rates for every one million tokens.

Large Language Models

  • DeepSeek V4.1 Flash
    Input / 1M
    0.20 €
    Cache / 1M
    0.10 €
    Output / 1M
    0.50 €
  • GLM 5.3 754B
    Input / 1M
    1.00 €
    Cache / 1M
    0.25 €
    Output / 1M
    3.50 €
  • GPT-OSS 120B
    Input / 1M
    0.15 €
    Cache / 1M
    —
    Output / 1M
    0.55 €
  • Llama 3.1 8B(EOL 30.09.26)
    Input / 1M
    0.15 €
    Cache / 1M
    —
    Output / 1M
    0.15 €
  • Llama 3.3 70B(EOL 30.09.26)
    Input / 1M
    0.65 €
    Cache / 1M
    —
    Output / 1M
    0.65 €
  • Mistral 4 119B
    Input / 1M
    0.20 €
    Cache / 1M
    —
    Output / 1M
    0.60 €
  • Qwen 3.8 27B
    Input / 1M
    0.30 €
    Cache / 1M
    0.10 €
    Output / 1M
    2.20 €

A few things to know.

What is OxVillage?

OxVillage is a Dutch AI company. Our inference service gives platforms, routers and developers a destination for requests to open models.

Where do models run?

Our inference offering runs on infrastructure in the Netherlands. If your workload has specific data-processing requirements, discuss them with us before sending production data.

Can I use an OpenAI-compatible client?

Yes. Configure your client with the API key, base URL and model identifier supplied with your access. Feature support and limits depend on the model and your configuration.

How is usage priced?

The inference page lists input, cached input where available, and output rates per million tokens. Contact us to discuss expected usage and any capacity requirements.

All questions & answers

Route your traffic to the
Dutch execution layer.

Join the routers, gateways and platforms running open models on Dutch infrastructure. One OpenAI-compatible API — built for your existing workflow.

Get API accessIntegration guide