Inference platform · NVIDIA GPUs

Open models, one API.

Galerna is building an inference platform for open-weight models: one OpenAI-compatible API on NVIDIA GPUs, per-token access through OpenRouter, and dedicated endpoints for teams that need capacity of their own.

Pre-launchOpenAI-compatibleEU-based · Madrid
Storm front over the Cantabrian Sea

01 · Models

Start with the open models teams already use.

A short list, served properly: every model runs at the precision we publish, and we never quietly serve a lower-precision version.

Qwen3.8-27BFirst model at launch
Tencent Hy3Next candidate
Nemotron 3.5 Super VLNext candidate · vision
Open Spanish modelPlanned · first stage
  • Served with vLLM on NVIDIA GPUs, through an OpenAI-compatible API.
  • Per token through OpenRouter, or on a dedicated endpoint reserved for you.
  • We’re EU-based. As of November 2026, none of these models has an EU-tagged endpoint on OpenRouter.
Waves breaking over a reef on the Cantabrian coast
The galerna: the sudden storm wind of Spain’s Cantabrian coast.

02 · Products

Pay per token, or reserve capacity.

Three ways to use the same models, from a first test to steady production traffic on GPUs reserved for you.

On OpenRouter

Not listed yet

Galerna as an inference provider on OpenRouter, starting with Qwen3.8-27B. Pay per token with your existing OpenRouter account, with no contract with us needed.

Pay per token · prices set at listing

Direct

Direct API

Planned

OpenAI-compatible API keys from us, with a contract, an invoice and a data processing agreement.

Reserved

Dedicated endpoints

Planned

A model on NVIDIA GPUs reserved for you and shared with no one, for steady production traffic.

About €2,500 per month per endpoint · planned price

  • OpenAI-compatible API
  • Published precision for every model
  • Zero retention of prompts and outputs
  • EU-based, on NVIDIA GPUs in Madrid

Tokens through OpenRouter, reserved capacity direct.Nothing is live yet. Planned prices, for information only; binding terms only in a signed contract.

03 · How it works

The model you ask for, at the precision we publish.

Your app
  • OpenAI SDK
  • OpenRouter
  • Any HTTP client
Galerna API
  • API keys
  • Usage metering
  • Zero retention
Model registry
  • Published precision
  • Unverified? Not served
NVIDIA GPUs
  • vLLM on CUDA
  • RTX PRO 6000 · 96 GB
Built

OpenAI/Ollama-compatible gateway, prefill and decode metering, and verification that fails closed. 84 automated tests; end-to-end smoke test against a simulated vLLM backend.

Next

First NVIDIA GPUs online, then OpenRouter’s provider process: application, review, an intro call, security questionnaire and terms, go-live. Billing and sign-up for direct customers.

04 · NVIDIA in our stack

Every Galerna model runs on NVIDIA GPUs.

We start on rented NVIDIA GPUs and add our own servers as reserved capacity sells.

Serve · first

Inference

Open-weight LLMs served with vLLM on CUDA, starting with Qwen3.8-27B. Starting card: NVIDIA RTX PRO 6000, 96 GB, rented.

Scale · planned

Owned servers

Our own NVIDIA servers once 60% of their hours are sold, growing into a small fleet in a colocation facility in Spain.

Verify · roadmap

Verified serving

Built in: precision checks before a model is listed. Next: hardware-rooted proof with NVIDIA confidential computing on Blackwell-class GPUs.

CUDAvLLMNVIDIA RTX PRO 6000 · 96 GBEvaluating: TensorRT-LLM · NIM · NeMoRoadmap: confidential computing

05 · Where we are

Nothing is live yet. Here’s the order we’re building in.

No GPUs running, no customers and no OpenRouter listing yet. We rent GPUs first and buy our own once their capacity is already sold.

Prove · Dec 2026 – Jun 2027

Serve on OpenRouter

  • Listed as an inference provider on OpenRouter
  • Open models on 1–2 rented NVIDIA GPUs
  • First direct pilots
  • One open Spanish model
Operate · Jul 2027 – Jun 2028

Dedicated endpoints

  • Dedicated endpoints for direct customers
  • First owned servers once 60% of their hours are sold
Own · Jul 2028 – 2029

Our own fleet

  • A small owned NVIDIA fleet in a colocation facility in Spain

06 · Company

The company behind Galerna.

Company

Galerna is a commercial brand of Unisphere Overseas S.L.

Based in

Madrid, Spain (EU)

What we do

Inference for open-weight models: OpenRouter, direct API, dedicated endpoints

Customers

Developers and companies building on open models

Stage

Pre-seed. API gateway built; first GPUs and OpenRouter listing next

Industry

Generative AI · AI inference

Contact

Build on open models.

We’re lining up our first pilot customers for the direct API and dedicated endpoints. Tell us which models you run and at what volume, and we’ll tell you when we can serve them.