Enterprise Pricing

Pick your model. We size and price the hardware.

A flat monthly subscription per dedicated node, tiered by GPU count and size. No per-token metering, no spot-price swings. Prices are indicative — talk to us to finalise.

public Dedicated in-region GPU · Switzerland available first

1× GPU

96 GB total VRAM
CHF 2,058/mo
indicative · per node

Single 96 GB GPU. Fits ~30B-class models comfortably. A strong starting node for a focused team.

Get started arrow_forward
Popular

2× GPU

192 GB total VRAM
CHF 4,075/mo
indicative · per node

192 GB total VRAM via tensor parallelism. Hosts large or long-context models such as ~120B configurations.

Get started arrow_forward

4× GPU

384 GB total VRAM
CHF 8,111/mo
indicative · per node

384 GB total VRAM — the largest single-node configuration. For the biggest models at maximum context.

Get started arrow_forward

Prices are indicative and shown in CHF per node per month. Final pricing depends on region, commitment, and configuration. Multi-GPU nodes use vLLM tensor parallelism across 96 GB-per-GPU cards. The largest open-weight flagships (for example GLM-5.2 or Kimi K2.7 Code at maximum context) can exceed a standard 4-GPU node — contact us to scope a custom configuration.

Which node size fits your model?

Pick an exemplary model to see which node sizes can host it. A size fits when its total VRAM (GPU count × 96 GB) clears the model's footprint plus KV-cache headroom.

~54 GB — a fast mid-size coder that fits a single GPU with room for context.

Best fit
1× GPU96 GB
CHF 2,058/mo
check Fits this model
2× GPU192 GB
CHF 4,075/mo
check Fits this model
4× GPU384 GB
CHF 8,111/mo
check Fits this model

Model footprints are rough, illustrative estimates and depend on quantization, context length, and concurrency. Prices are indicative CHF per node per month.

Every node includes

  • checkDedicated, single-tenant, always-on node in your chosen region
  • checkPrompts and completions never leave the node — telemetry-only uplink
  • checkDirect routing with on-node TLS and on-node key authentication
  • checkFrontier open-weight models (MiniMax M3, GLM-5.2, Kimi K2.7 Code)
  • checkOpenAI-compatible /v1/chat/completions endpoint
  • checkOrganisation workspace: nodes, members, API keys, subscription
  • checkFlat monthly subscription — no per-token metering, no overage

Pricing FAQ

Ready for a Dedicated In-Region Node?

Start an enterprise workspace to provision your first node, or contact us to finalise pricing for your region and configuration.

Start an Enterprise Workspace arrow_forward Data residency & trust