40+ open-source models, every request processed in EU data centres, zero data retention. Drop-in OpenAI replacement.
Top up credits and receive your API key instantly. No lengthy onboarding, no sales calls, no enterprise minimum.
Same OpenAI format you already use. Change the base URL and key. Your existing code works without modification.
Pay per token. No monthly minimum, no per-seat charges, no surprises.
Hosted across Finland, France, Sweden. No US routing, no China routing. Prices shown in USD per 1M tokens — Router automatically picks the cheapest EU host for each call.
| Model | Region | Input / 1M | Output / 1M | Best for |
|---|---|---|---|---|
| mistral-nemo | 🇫🇷 | $0.02 | $0.02 | Long-context Mistral |
| gpt-oss-20b | 🇫🇮 🇫🇷 | $0.05 | $0.17 | Cheapest. Background tasks. |
| nemotron-nano-30b | 🇫🇮 | $0.07 | $0.28 | Fast routing, extraction |
| qwen3-coder-30b | 🇫🇷 | $0.07 | $0.25 | Code generation |
| gpt-oss-120b | 🇫🇮 🇫🇷 | $0.09 | $0.46 | General purpose, tool use |
| qwen3-32b | 🇫🇮 🇫🇷 | $0.09 | $0.26 | General purpose |
| mistral-small-3.2 | 🇫🇷 | $0.10 | $0.32 | Vision + text |
| mistral-7b | 🇫🇷 | $0.11 | $0.11 | Compact, fast |
| qwen3-30b | 🇫🇮 | $0.11 | $0.34 | Chat, reasoning |
| llama-3.3-70b | 🇫🇮 🇫🇷 🇸🇪 | $0.14 | $0.44 | Reliable all-rounder |
| hermes-4-70b | 🇫🇮 | $0.15 | $0.46 | Memory, learning agents |
| qwen3-next-80b-thinking | 🇫🇮 | $0.17 | $1.38 | Deep reasoning, thinking |
| intellect-3 | 🇫🇮 | $0.23 | $1.26 | Open reasoning |
| llama-3.1-8b | 🇫🇷 | $0.23 | $0.23 | Edge, low-latency |
| mistral-small | 🇫🇷 | $0.23 | $0.69 | Balanced reasoning |
| qwen3-235b | 🇫🇮 🇫🇷 | $0.23 | $0.69 | Complex reasoning, long context |
| gemma-3-27b | 🇫🇷 | $0.29 | $0.57 | Compact Google open model |
| minimax-m2.5 | 🇸🇪 | $0.32 | $1.26 | MoE, 200k context |
| deepseek-v3.2 | 🇫🇮 | $0.34 | $0.52 | Efficient reasoning + code |
| holo2-30b | 🇫🇷 | $0.34 | $0.80 | Conversational AI |
| devstral-2-123b | 🇫🇷 | $0.46 | $2.30 | Large code model |
| kimi-k2 | 🇫🇮 | $0.57 | $2.76 | Agentic, tool orchestration |
| kimi-k2.5 | 🇸🇪 | $0.57 | $2.76 | Agentic, 256k context |
| glm-4.5 | 🇫🇮 | $0.69 | $2.53 | Bilingual Chinese-English |
| qwen3.5-397b | 🇫🇷 | $0.69 | $4.14 | Frontier multimodal MoE |
| deepseek-r1-distill-70b | 🇫🇷 | $1.03 | $1.03 | Reasoning distill |
| hermes-4-405b | 🇫🇮 | $1.15 | $3.45 | Maximum capability |
| glm-5.1 | 🇸🇪 | $1.61 | $5.06 | Long-context multilingual |
| mistral-large | 🇫🇷 | $2.30 | $6.90 | Frontier Mistral |
Plus 5 embedding models, 2 vision models, 1 speech-to-text models, 2 image generators, and 1 safety classifier(s). Full list via the /v1/models endpoint.
Every request processed in Finland, France, Sweden. No US routing. No China routing. Architecturally guaranteed.
Inputs and outputs are never stored, logged for training, or routed to non-EU systems. Vokt holds the upstream contracts so you only sign one DPA — with us.
Built for European regulation, not retrofitted. SOC 2, HIPAA, ISO 27001 certified infrastructure. Custom DPAs available.
Provider down? Transparent switch to the next EU provider. Same model, different infrastructure. No downtime for your application.
When the same model is available on multiple providers, we route to the cheapest one. You get the best EU price automatically.
Same API format. Change two lines in your code. Works with every OpenAI SDK, LangChain, LlamaIndex, and any tool that supports custom endpoints.
No monthly minimum. No per-seat fees. No hidden costs. Buy credits, use any model, pay per token.
One API key. 40+ models. Every request stays in Europe. Start in five minutes.
Get Started