Pricing

Pay the provider's rate, plus a flat 8%.

One usage-based gateway for every model you run. Metered per token, billed through Stripe, and PHI routing carries no surcharge. No seats, no minimums — the bill scales only with what you use.

Provider rate + 8%No PHI surcharge$0 self-hosted tier
phi-cloud · CH
$ POST /v1/chat/completions
  X-PHI: true   X-Region: CH

< HTTP/2 200
< x-phi-tier: phi
< x-phi-routed: infomaniak/CH
< x-phi-attempts: 1

How pricing works

Two numbers, and nothing hidden behind them.

The whole model is a pass-through rate and a flat margin. Here is every part of it, with nothing you have to read the fine print to find.

Pass-through

The provider rate, untouched

You pay each upstream model at its own per-token rate. We don't mark up the model — the rate you see is the rate the provider charges, billed per token.

Margin

A flat 8% gateway fee

On top of the provider rate, the gateway adds a flat 8% — the only thing phi-cloud charges. No tiers, no enterprise upcharge, the same 8% at every volume.

PHI included

No surcharge to route PHI

Pinning a request to a PHI-eligible provider costs nothing extra. The Swiss PHI tier is priced exactly like any other route — privacy is not a premium.

Stateless

Pay per call, nothing stored

Cost is computed on each response and metered into Stripe. There is no application database and no payload retention — you are billed for tokens, never for storage.

No lock-in

No seats. No minimums.

There are no per-user licenses and no monthly floor. The bill scales with usage and falls to zero when you stop calling. Auto top-up keeps you running.

Private by default

The self-hosted tier is $0

General (non-PHI) traffic runs on the self-hosted Medishift worker at $0 per token. You only reach metered provider rates when you opt into a hosted model.

A worked example

1M in + 1M out on the Swiss PHI tier

Send one million input tokens and receive one million output tokens through Mistral 24B (Infomaniak, CH). At $0.30 in and $0.90 out per million, the whole run lands at the price below — gateway margin already included, PHI routing at no extra cost.

1M input tokens
$0.30
1M output tokens
$0.90
Total
$1.20

Live model prices

Every model, every rate, on one page.

Listed per 1M tokens, margin included. Regional rows reflect the local provider price — often 5–10× cheaper than US frontier models. PHI-eligible routes are highlighted.

ModelRegionPHIIn / 1MOut / 1M
Gemma-4 E2B IT (Medishift worker)
worker-gemma-4-e2b
multigeneral$0.00$0.00
Qwen3-Embedding-8B · 4096-d (Medishift worker)
worker-qwen3-embedding-8b
multigeneral$0.00$0.00
Mistral 24B (Infomaniak, CH)
infomaniak-mistral-24b
CHeligible$0.30$0.90
Mistral 3 (Infomaniak, CH)
infomaniak-mistral-3
CHeligible$0.20$0.60
Qwen3 (Infomaniak, CH)
infomaniak-qwen3
CHeligible$0.30$0.90
BGE-Multilingual-Gemma2 · 3584-d (Infomaniak, CH)
infomaniak-bge-gemma2
CHeligible$0.05$0.00
All-MiniLM-L12-v2 · 384-d (Infomaniak, CH)
infomaniak-minilm-l12-v2
CHeligible$0.02$0.00
Claude Opus 4.7
claude-opus-4-7
multigeneral$15.00$75.00
Claude Sonnet 4.6
claude-sonnet-4-6
multigeneral$3.00$15.00
Claude Haiku 4.5
claude-haiku-4-5
multigeneral$1.00$5.00
GPT-5
gpt-5
multigeneral$5.00$15.00
GPT-5 mini
gpt-5-mini
multigeneral$0.25$2.00
Gemini 2.5 Pro
gemini-2.5-pro
multigeneral$2.00$10.00
Mistral Large 2
mistral-large-2
EUgeneral$2.00$6.00
Mistral Small
mistral-small
EUgeneral$0.20$0.60
Aleph Luminous
aleph-luminous
EUgeneral$3.00$9.00
Qwen Max
qwen-max
CHINAgeneral$0.40$1.20
Qwen Plus
qwen-plus
CHINAgeneral$0.10$0.30
ERNIE 4.5
ernie-4-5
CHINAgeneral$0.50$1.50
GLM-4 Plus
glm-4-plus
CHINAgeneral$0.30$1.00
Kimi K2
kimi-k2
CHINAgeneral$0.20$0.80
MiniMax abab-7
minimax-abab-7
CHINAgeneral$0.15$0.60
Claude Sonnet 4.6 (Tokyo)
claude-sonnet-4-6-jp
ASIAgeneral$3.00$15.00
Mistral Large 2 (Singapore)
mistral-large-sg
ASIAgeneral$2.00$6.00
Jais 70B (UAE)
jais-70b
MENAgeneral$0.80$2.40
Sarvam-1
sarvam-1
INDIAgeneral$0.30$1.00
Krutrim Spectre
krutrim-spectre
INDIAgeneral$0.25$0.80

Per 1M tokens, gateway margin included. The same rows are served live at GET /v1/models. The $0 rows are the self-hosted Medishift worker (non-PHI tier).

FAQ

Questions a compliance team asks.

The short, honest answers — the kind that survive a security review.

For PHI, yes — and phi-cloud's PHI tier runs on Infomaniak (CH) under a filed Swiss nFADP DPA. A signed customer BAA process is on the roadmap; today the PHI guarantee is the router pinning PHI to the eligible Swiss provider. Ask us about current BAA status for your jurisdiction.

Still have a question? Read the docs or start building.

Put AI in production — without giving up your data.

One private cloud for the models you already use. Your data stays in your region, never trained on, every call yours alone.