Obsidian Peaks
Pilot

Why Obsidian Peaks

Your own RTX 5090, the same machine every month. Owned by us and run in the Stockholm region, from a Swedish company.

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 open model on Hugging Face, pulled when the box starts.

  • Root access
  • Yours alone during your term
  • Disk wiped at teardown

Open models

Sweden

In the Stockholm region, by a Swedish company.

Our RTX 5090s are owned by us and run in the Stockholm region, Sweden. Public traffic exits through our server in Helsinki (EU). The contract, the invoice and the support come from Obsidian Peaks Technologies AB in Stockholm, under Swedish law.

  • Owned hardware, Stockholm region
  • Swedish law
  • Support in English and Swedish

Company

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

Reserve

Per token or GPU

Why the Peaks are 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 vs your own GPUPer-token APIYour own GPU
What you buyTokensThe GPU
Who else runs on the hardwareA shared poolNobody during your term
ModelsTheir shelfAny
AccessAPI onlyRoot + API on your box
PricingPer token, prepaid plan requiredPer month, published
When it winsSmall or bursty volumeSteady volume, isolation, custom stacks, fine-tuning

Full comparison

Break-even

Where the line goes.

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.

Where their bill passes the boxKimi-K3 passes the box at 48 M tokens. Qwen3.8-27B-FP8 passes the box at 242 M tokens. GLM-5.3-Flash · Gemma-4-31B passes the box at 1.5 B tokens. Mistral-Small-3.2-24B passes the box at 2.4 B tokens08 00016 00024 00001.0 B2.0 B3.0 BOutput tokens a monthSEK a monthKimi-K3Qwen3.8-27B-FP8GLM-5.3-Flash · Gemma-4-31BMistral-Small-3.2-24BObsidian Peaks, 8 000 SEK48 M242 M1.5 B2.4 B
Their list prices for output tokens, in SEK; the box at its monthly price. The dots are where a month of tokens costs more than the box.

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.

Tokens

A month, priced both ways.

What one box makes in a month at full load, and what that month costs at their price.

A month of one box, priced at their API
ModelA month at full load, one boxThat month at BergetObsidian Peaks
Kimi-K3about 20 boxes of memory, in one chassis · €3.00 · €15.00 per M tokens——on request
Qwen3.8-27B-FP8one box, tight fit · €0.40 · €3.00 per M tokens1.3 B tokens (~500/s)about 43 000 SEK8 000 SEKtwo boxes, room for context: 16 000 SEK
GLM-5.3-Flashsize not public · €0.25 · €0.50 per M tokens———
Gemma-4-31Bone box, 4-bit · €0.25 · €0.50 per M tokens2.6 B tokens (~1 000/s)about 14 000 SEK8 000 SEKtwo boxes, FP8: 16 000 SEK
Mistral-Small-3.2-24Bone box · €0.30 · €0.30 per M tokens2.6 B tokens (~1 000/s)about 8 600 SEK8 000 SEK
KB-Whisper (speech-to-text)one box · €0.12 per hour of audio43 200 hours of audio (~60× realtime)about 56 000 SEK8 000 SEK
E5 embeddingsone box · €0.03 per M tokens51.8 B tokens (~20 000/s)about 17 000 SEK8 000 SEK
FLUX.1 (image generation)one box · not on their list1.0 M images (~24 a minute)not on their list8 000 SEK

The month’s output is our estimate for one 5090 kept busy around the clock, output tokens only; measured figures from our own boxes will replace it. Send less than that and their API is cheaper on price alone. Send more, or want what the box adds, and the box wins.

Their prices: api.berget.ai model list, 23 Sep 2026; 11 SEK per euro; bought through a prepaid plan from €25 a month, credits non-refundable, usage past the credit at list price. Kimi-K3 is a trillion-parameter-class model: about 600 GB at 4-bit, twenty boxes’ worth of memory in one chassis; for Kimi, their API is the sane buy.

On top

What the box adds.

What a GPU of your own adds that no token can.

  • Yours alone

    One tenant per GPU. Prompts, files and weights stay on your box; traffic to it passes through our relay in Helsinki (EU).

  • Any model

    Text, vision, speech, image generation, your own fine-tunes: whatever fits in 32 GB.

  • Agents that never stop

    Coding agents and pipelines in Docker, around the clock, no meter. Their coding seats are €150–500 per person a month, rate-limited; one box serves the whole team.

  • Root access

    SSH in, install what you like, keep the state between jobs.

  • One price

    No per-token bill, no rate limit, no prepaid plan. Locked for your term.

Questions

Questions and answers.

What do I get when I rent a GPU?
One whole machine with an NVIDIA RTX 5090 (32 GB of VRAM), owned by us and run in the Stockholm region, Sweden: yours by the month under a signed agreement, the same machine every month, with root access.
How does billing work?
A published monthly price: 8 000 SEK/month ex VAT, or 7 200 SEK/month on a 12-month term, for companies. Storage, traffic and electricity are included, and the invoice comes from Obsidian Peaks Technologies AB.
What does “only you” mean?
During your agreement nobody else runs on the machine: it is yours alone. 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, pulled when the box starts. CUDA: what runs on an RTX 5090 runs here.
Where is the hardware, and whose company is it?
Our RTX 5090s are owned by us and run in the Stockholm region, Sweden. Public traffic exits through our server in Helsinki (EU). The contract, the invoice and the support come from Obsidian Peaks Technologies AB in Stockholm, under Swedish law.

Rent the GPU.

By the month, under a signed agreement.