How does serverless routing work?

Cortecs uses a filter-and-rank approach. Providers that don’t meet the request’s specific requirements are filtered out first. The remaining options are then ranked based on the selected inference preferences.

If a provider becomes unavailable, the request is automatically rerouted to the next best option to maintain uninterrupted service. For more details, see the docs.

How does it compare to OpenRouter?

Cortecs’ serverless inference works similarly to OpenRouter by automatically routing requests to the best available model based on demand. However, Cortecs is built with a strict focus on EU compliance. All data is processed exclusively within European jurisdictions

Can I use models on Hugging Face?

Yes, we support hosting Hugging Face models as dedicated deployments. This means you can run any compatible model from Hugging Face’s model hub on infrastructure provisioned specifically for your workloads.

Just post your request in our Discord channel.

Do you support RAG?

Cortecs is an excellent foundation for building data-sensitive Retrieval-Augmented Generation (RAG) applications. Whether you need a hybrid deployment with your data stored on-premises or hosted in a sovereign cloud.

Sky Inference handles your AI requests while your sensitive data stays where you want it.

Is my data stored or used for training?

No, Cortecs does not store your data or use it for model training. For more information see our Privacy Policy.