Why Obsidian Peaks
A GPU in Sweden that only you use. RTX 5090 by the hour or the month, from a Swedish company, on hardware we run ourselves in Sweden, billed by the minute.
The GPU
The GPU itself.
Root access on hardware nobody else uses during your term. CUDA, so what runs on an RTX 5090 runs here. Any model on Hugging Face, or one of the open models on our mirror, pulled when the instance starts.
- Root access
- Yours alone during your term
- Disk wiped at teardown
Price
A price you can check.
On demand: the RTX 5090 market median minus 15%, billed by the minute, locked 7 days at a time. Dedicated: a published monthly price. Your balance stays yours: unused balance is refunded on request, no credits that expire.
- Market median −15%
- Billed by the minute
- Unused balance refunded
Sweden
In Sweden, by a Swedish company.
The hardware is ours and it is in Sweden. The contract, the invoice and the support come from Obsidian Peaks Technologies AB in Stockholm, under Swedish law.
- Hardware in Sweden
- Swedish law
- Support in English and Swedish
Start
Running in minutes.
Open an account, add your SSH key, pick a box, connect. The Console opens VS Code, Cursor or Claude Code on your instance with one click.
- Sign up
- Add a key
- Pick a box
- Connect
- VS Code
- Cursor
- Claude Code
Dedicated
Dedicated inference box.
Your model, your OpenAI-compatible URL, on an RTX 5090 that is yours by the month: 8 000 SEK/month ex VAT, 7 200 SEK/month on a 12-month term.
- OpenAI-compatible endpoint (vLLM)
- Any open model
- Root access
Per token or GPU
Why Obsidian takes you higher than Berget
Berget AI sells a per-token API; we rent you the GPU.
| Per-token API vs your own GPU | Per-token API | Your own GPU |
|---|---|---|
| What you buy | Tokens | The GPU |
| Who else runs on the hardware | A shared pool | Nobody during your term |
| Models | Their shelf | Any |
| Access | API only | Root + API on your box |
| Pricing | Per token, prepaid plan required | Per minute, published |
| When it wins | Small or bursty volume | Steady volume, isolation, custom stacks, fine-tuning |
One dedicated 5090 at 8 000 SEK/month costs about the same as 1.5 billion output tokens at a typical Swedish per-token API’s list price. Below that, tokens are cheaper. Above it, or when you need isolation, a custom stack or fine-tuning, the GPU wins.
List price used: €0.50 per million output tokens, the published price of a 31B open model at a Swedish per-token API on 23 Sep 2026; 8 000 SEK/month converted at about 11 SEK per euro. Ex VAT.
Questions
Questions and answers.
- What do I get when I rent a GPU?
- One NVIDIA RTX 5090 with 32 GB of VRAM on a machine we run in Sweden, with root access and the disk size you chose. On demand it is a container with exclusive use of the GPU; dedicated it is the whole machine.
- How does billing work?
- On demand: a prepaid balance, GPU time billed by the minute while the instance runs, disk for as long as the instance exists and traffic as used; the price is locked 7 days at a time and follows the RTX 5090 market median minus 15%. Dedicated: a published monthly price, 8 000 SEK/month ex VAT, with storage and traffic included.
- What does “only you” mean?
- During your term nobody else runs on the hardware you rent: on demand the GPU is yours alone, dedicated the whole machine is. We keep an operations key for support, restores, urgent security maintenance and incidents; every use is logged and we tell you afterwards.
- Which models can I run?
- Any open model on Hugging Face, or one of the open models on our mirror, pulled when the instance starts. CUDA: what runs on an RTX 5090 runs here.
- Where is the hardware, and whose company is it?
- The machines are our own and they are in Sweden. The contract, the invoice and the support come from Obsidian Peaks Technologies AB in Stockholm, under Swedish law, and your data is not moved out of the EU/EEA.
- How fast am I connected?
- Open an account, add your SSH key, pick a box, connect: an on-demand instance is running in minutes, and the Console opens VS Code, Cursor or Claude Code on it with one click.
- Can I cancel, and what happens at teardown?
- You stop or destroy an instance yourself at any time. Usage you have consumed is not refundable; unused balance is refunded to the original payment method on request. Destroying an instance deletes its disk permanently; we keep no backups, so keep your own copies.