Solutions
From one API call to your own GPU cluster.
Inference, agent sandboxes, and public cloud in one portal, and dedicated NVIDIA GPUs when you need your own. Hosted in Europe.
Open-weight models through an OpenAI-compatible API.
DeepSeek, Kimi, and GLM behind one endpoint and one key, billed per token. Swap your base URL and key, and your OpenAI code keeps working.
Explore InferenceFAQ
Questions, answered.
Can’t find what you need? Our engineers are happy to walk through your workload.
No. Start a 30-day free trial with object storage and inference access, with no payment method required, then activate the usage-based plan in the portal. Dedicated capacity is arranged with our team.
Currently, entirely in Europe. Every self-serve service, inference included, runs on infrastructure in the European Union, and dedicated deployments can also run in Norway by agreement. If new regions are added, using them will be your choice: your data stays where it is unless you decide to move it.
No. Prompts and outputs are processed transiently and never stored, logged, or used to train models. We record only metadata such as model, token counts, and timestamps for billing.
Compute, including networking, is in Early Access. It has no availability commitment yet.
Buckets are encrypted with managed keys, one per bucket, chosen at creation. Customer-managed keys are available by arrangement with support.
No. Storage speaks S3 and inference speaks the OpenAI API. There are no switching or early-termination charges, and egress is free when you move.
Open the portal. See it for yourself.
Create an account, make a bucket, and send your first inference call in minutes. No sales call, no commitment.
30-day free trial · no payment method required