Transparent pricing. No percentage of your tokens.
Provider costs pass through at cost — Trifrost charges a flat platform fee, never a markup on upstream model pricing. Percentage-markup routers make more when you spend more. We don't. Identical pricing across all five regions.
Free
$0
- ▸ 1M tokens/mo gateway throughput (hard cap)
- ▸ Shared default region
- ▸ 1 user · 1 department · 1 virtual key
- ▸ Playground + basic spend view
- ▸ Community support
Team
$399/mo
or $3,990/yr
- ▸ 25M tokens/mo included · $2 per extra 1M
- ▸ 1 region — your pick
- ▸ 10 users · 3 departments · 25 virtual keys
- ▸ Per-model budget matrix
- ▸ Spend analytics + CSV/JSON export
- ▸ Playground · email support
Business
$1,499/mo
or $14,990/yr
- ▸ 150M tokens/mo included · $2 per extra 1M
- ▸ 1 region — your pick
- ▸ Unlimited users, departments, virtual keys
- ▸ Model allowlists + session cost attribution
- ▸ Spend forecast + alerts
- ▸ Audit export (JSON/JSONL/Syslog)
- ▸ In-flight observability · priority support
Enterprise
Contact
custom agreement
- ▸ Unlimited throughput
- ▸ Region pinning — AU·APAC·EMEA·UK·DE, or we stand up a new region
- ▸ SSO/SAML + custom OIDC
- ▸ DPA + sub-processors review
- ▸ 24/7 support + named CSM
- ▸ Roadmap influence
Provider costs always passed through at cost, on every tier. All prices USD.
Every feature, every tier.
| Feature | Free | Team | Business | Enterprise |
|---|---|---|---|---|
| Gateway throughput | 1M tokens/mo | 25M tokens/mo | 150M tokens/mo | Unlimited |
| Overage | Hard cap | $2 / 1M tokens | $2 / 1M tokens | — |
| Provider costs | At cost | At cost | At cost | At cost |
| Region | Shared default | 1 region, your pick | 1 region, your pick | Pinned · any of 5 + region on request |
| Users | 1 | 10 | Unlimited | Unlimited |
| Departments | 1 | 3 | Unlimited | Unlimited |
| Virtual keys | 1 | 25 | Unlimited | Unlimited |
| Per-model budget matrix | — | ✓ | ✓ | ✓ |
| Model allowlists | — | — | ✓ | ✓ |
| Spend analytics + export | Basic view | ✓ | ✓ | ✓ |
| Session-level cost attribution | — | — | ✓ | ✓ |
| Spend forecast + alerts | — | — | ✓ | ✓ |
| Audit log export (JSON/JSONL/Syslog) | — | — | ✓ | ✓ |
| In-flight observability | — | — | ✓ | ✓ |
| Playground | ✓ | ✓ | ✓ | ✓ |
| SSO/SAML + custom OIDC | — | — | — | ✓ |
| DPA + sub-processors review | — | — | — | ✓ |
| Support | Community | Priority | 24/7 + named CSM |
What does one outage cost you?
The platform fee is the small number. These are the big ones.
A provider outage mid-quarter-end
Alias-group failover keeps your apps up when a provider blips — including zero-content stream truncation, where the provider returns 200 and dies anyway. Your users never see it.
An untracked intern calling Opus all month
Per-model budgets and allowlists cap it at the key, the user, or the department — enforced pre-upstream with a structured 402, not discovered on the invoice.
A regulator asks who called what
The HMAC-signed event log answers in one query — every request, including the blocked ones, with tamper verification on read. No forensics project, no guesswork.
Rebuilding provider integrations quarterly
27+ provider kinds behind one maintained, OpenAI-compatible API. Vendors change endpoints, pricing, and reliability — your code doesn't.
Pricing questions, straight answers.
Is there a markup on tokens?
No. Provider costs pass through at cost — you pay your providers' published rates, plus a flat Trifrost platform fee per tier. We never take a percentage of your token spend, so our revenue doesn't grow when your bill does.
What counts as gateway throughput?
Prompt + completion tokens proxied through the gateway, metered per request. Cache read/write tokens are counted separately at their tiered rates.
What happens when I hit my cap?
Free hard-caps at 1M tokens/month. Team and Business keep running — overage bills at $2 per additional 1M tokens. Budgets and spend alerts mean you see it coming long before it matters.
Can I choose my region?
Team and Business pick one region at signup — AU, APAC, EMEA, UK, or Germany, all live. Enterprise pins any of the five, and we'll stand up a new region on request.
Is pricing different by region?
No. Pricing is identical across all five regions. You pick a region for latency and residency, not cost.
Do you offer annual billing?
Yes — Team is $3,990/yr and Business is $14,990/yr, roughly two months free against monthly billing. Enterprise agreements are custom.
Your first million tokens are on us.
Deployed in AU · APAC · EMEA · UK · DE. Same price everywhere. Need somewhere else? We'll stand it up.