Solheim

Solheim rents dedicated, self-hosted large language model instances located in the EU, billed as a fixed monthly fee rather than per token. Each instance is described as OpenAI-compatible, so it can be pointed at existing tools such as Cline or VS Code extensions that expect an OpenAI-style API without code changes. The service is labeled 'Issue 01 — a VPS for language models' on its site, framing it as an early-stage offering. It targets developers and teams who want predictable LLM costs and EU-based hosting — useful for data-residency requirements or workloads with steady, high-volume usage where per-token pricing becomes expensive — rather than casual or bursty use cases better served by pay-as-you-go APIs. Pricing starts at €15 per month for one concurrent instance with a 64k context window, scaling with instance count and context size; the site states there are 'no tokens, no tiers and no overage,' with larger enterprise deployments priced against the underlying hardware. One limitation: because capacity is dedicated rather than elastic, the site says the only real constraint is how many simultaneous requests run against an instance at once, meaning traffic spikes beyond a rented instance's capacity would need an upgrade rather than automatically scaling.