Pricing · one page, every list price
Dedicated GB10 compute per hour: on demand, scheduled, or cheaper by the week, month or year. Serverless models per token, image, minute or track. Pay-as-you-go, no subscription, no minimum, no hidden tiers.
Dedicated GPU rental
Dedicated machine, 128 GB unified memory, full SSH access, your own environment. Available
| Product | What you get | Region | Price |
|---|---|---|---|
| On demand | Pay per started hour you use, and switch it on and off in the portal | eu-es-1 · Málaga | €1.19 / hour |
| Scheduled | The exact hours you book on the calendar | eu-es-1 · Málaga | €0.99 / hour |
| 1 week | Scheduled hours, booked a week or longer | eu-es-1 · Málaga | €0.95 / hour |
| 1 month | Scheduled hours, booked 28 days or longer | eu-es-1 · Málaga | €0.91 / hour |
| 1 year | Scheduled hours, booked a year | eu-es-1 · Málaga | €0.85 / hour |
Longer reservations receive a lower effective compute rate. List prices in EUR. Trial access is available for qualifying customers. Check availability or Reserve capacity in the console.
| Other GPUs | Memory | Price | Status | Next step |
|---|---|---|---|---|
| RTX 3060 ×2: dual-GPU machine (starter tier), or one card | 12 GB each | €0.41 / hour · one card €0.24 | Available | Check availability |
| RTX 6000 Pro | 96 GB | €1.58 / GPU-hr | Request capacity | Ask for this GPU |
| RTX 5090 · 4-GPU system | 32 GB each | €1.26 / GPU-hr | Available | Ask for this GPU |
| RTX 3090 · 4-GPU system | 24 GB each | €0.62 / GPU-hr | Request capacity | Ask for this GPU |
| RTX 4080 | 16 GB | — | Request capacity | Ask for this GPU |
| H100 / H200 / B200 | 80 / 141 / 192 GB | — | Request capacity | Ask for this GPU |
Request-capacity systems carry their launch price per GPU-hour; the deployment is confirmed with your request, and registered demand is used to prioritise new capacity. List prices in EUR.
Serverless Models
3M free tokens every 30 days with every new account.
| Model | Unit | Price | Region | Status |
|---|---|---|---|---|
| Qwen3.8 27B: input | 1M tokens | €0.32 | eu-se-1 · Stockholm | Available |
| Qwen3.8 27B: output | 1M tokens | €2.00 | eu-se-1 · Stockholm | Available |
| Xev decisions: input (output is free) | 1M tokens | €0.04 | eu-se-1 · Stockholm | Available |
| Qwen3 Embedding | 1M tokens | €0.02 | eu-se-1 · Stockholm | Available |
| ERNIE Image Turbo | image | €0.05 | eu-se-1 · Stockholm | Available |
| FLUX.2 Klein 4B edits | edit | €0.06 | eu-se-1 · Stockholm | Available |
| Whisper | minute | €0.005 | eu-se-1 · Stockholm | Available |
| Piper | 1M characters | €2.95 | eu-se-1 · Stockholm | Available |
| MiniMax Music 3 | 10 seconds | €0.04 | eu-se-1 · Stockholm | Available |
Pay-as-you-go, no subscription, no minimum. Every API response reports token counts. List prices in EUR.
Managed GPU
| Deployment | Hardware from | Managed service | Region |
|---|---|---|---|
| NVIDIA DGX Spark (GB10), 128 GB unified, operated for you | €664.30 / month | Quoted per deployment | eu-es-1 · Málaga |
| Additional GPU capacity: RTX 6000 Pro, RTX 5090, RTX 3090, H100, H200, B200 | Against registered demand | Quoted per deployment | EU |
| Model deployment: validated open models or your own weights | — | Quoted per deployment | EU |
Deployment and operation included: machine, CUDA, runtime, endpoint. The hardware price and the managed-service fee are confirmed in writing before anything is billed. Request deployment.
GPU VM
The self-managed form of the dedicated rental above: full SSH access, your own containers and software stack, priced per hour and cheaper the longer you book: €1.19/hour on demand · €0.95/hour by the week · €0.91/hour by the month · €0.85/hour by the year for the DGX Spark (GB10); Dual RTX 3060 machines from €0.29/hour. Additional GPU classes on request. Check availability.
How we price
Every price is on this page, in euros, excluding VAT. You pay for what you use, with no subscription and no minimum. Committed volume? Talk to us.
Why so cheap
Three reasons, none of them a trick:
| We own the hardware | Inference runs on NVIDIA DGX Spark systems AxForge owns and operates in Sweden, with no cloud markup between you and the silicon. |
|---|---|
| Efficient models, right-sized machines | We serve efficient open-weight models on machines sized for exactly that job, so the hardware is actually utilized instead of idling expensively. |
| Real cost plus a thin margin | The price reflects what serving actually costs us, plus a thin margin. No VC-subsidized loss-leader pricing that has to snap back later. |
What the low price does not buy us: your data. Prompts and completions are processed in memory, never retained, never used to train. See the privacy policy.
FAQ
Every new account starts with a free monthly token allowance, no card required. Pay-as-you-go starts at the listed per-token prices, with no subscription and no minimum.
These are the opening prices for the shared API. It is not a teaser with a preset expiry: when a price changes, this page changes with it.
They can, in both directions, and this page always shows the current price. What never changes with the price: EU residency, zero prompt retention, and no training on customer data.
Chat is billed per token with separate input and output rates; embeddings per input token; image generation per image. Every API response reports usage, including in streaming mode, so you can reconcile your bill exactly. List prices in EUR.
There are no published volume tiers. For sustained high volume, a dedicated system, a monthly DGX Spark rental serving only your traffic, is usually the better shape. Talk to an engineer and we’ll quote it.
Related