Pricing
You pay the token price of the model you call, plus one visible platform fee. Nothing is added to the provider's rate in secret.
Card payments are not enabled on this deployment, and nothing adds credit automatically: there is no invoice run and no operator credit path yet. The console shows what you have spent and what remains.
This deployment may run mock inference: a response can come from a stand-in model rather than the model named in the request. When mock mode is on, the console and the trace say so, and no token charge is made.
Models
| Models | Region | in per million tokens | out per million tokens |
|---|---|---|---|
| Whisper large-v3 | de-fra | 0,0736 € | 0,22 € |
| Mistral Small 4 (119B-A6B) | fr-par | 0,0920 € | 0,28 € |
| EuroLLM-1.7B-Instruct | de-fra | 0,0920 € | 0,28 € |
| EuroLLM-22B-Instruct (2512) | de-ber | 0,11 € | 0,33 € |
| Devstral Small 2 (24B) | de-ber | 0,32 € | 0,97 € |
| EuroLLM-9B-Instruct (2512) | de-fra | 0,32 € | 0,97 € |
6 endpoint(s) currently priceable. If a model costs more through LLM EU than directly from its provider, the card says so and by how much. There is no hidden markup.
What the platform fee pays for
The router and its policy engine, the EU host and its operations, the traces you read in the console, and the work of documenting cards. It is charged once per request, shown on every invoice line, and never folded into a token price.
If a model costs more through LLM EU than directly from its provider, the card says so and by how much. There is no hidden markup.
Enterprise
For teams assessing policy-controlled inference and auditable request metadata. Review the implemented controls, current limitations and unfinished contractual documents before considering production use.