Skip to content
Enterprise AI gateway

One governed API for every model.

Trifrost is the enterprise control plane for AI inference — load balancing, failover, cost control, and audit across 27+ providers, behind your identity provider, in the region you need it.

Book a demo See it fail over →
AU · APAC · EMEA · UK · DE SOC 2 Type II ISO 27001 HIPAA
The Trifrost playground: a kimi3 chat session with reasoning bubbles, per-message token and cost telemetry, and a session cost rollup of $0.0052 in the header
The playground — every turn billed, metered, and attributed
Why "Trifrost"

Three bridges in one.

Bifrost is the Norse bridge between realms. Trifrost is the three bridges your organization actually needs to run AI in production.

Providers → You

OpenAI, Anthropic, Bedrock, Vertex, OpenRouter, Fireworks, Together, DigitalOcean, Ollama — 27+ provider kinds collapse to one stable, OpenAI-compatible API. Your apps never rewire when a vendor changes pricing, endpoints, or reliability.

You → Your teams

One gateway, many departments and users. Each team gets its own virtual keys, per-model budgets, rate limits, and allowed models — mapped from your IdP groups, not managed by hand.

Teams → Compliance

Every call is attributed, costed, and written to a tamper-evident, HMAC-signed event log. Regulators and finance get the answers they need; engineers keep moving fast.

The platform

Gateway. Governance. Proof.

Gateway

27+ providers

  • ▸ Alias-group load balancing across providers
  • ▸ Zero-content stream failover — sessions survive a provider cutting out mid-stream
  • ▸ Per-request provider directives: only / ignore / order
Explore the gateway →

Governance

zero DB hits on the hot path

  • ▸ Per-model budgets per department, user, and key
  • ▸ Model allowlists — regulatory model restriction by team
  • ▸ Session-level cost attribution, not just per-call
Explore governance →

Observability

1 request = 1 signed event

  • ▸ HMAC-signed, tamper-evident audit log — always on
  • ▸ Live in-flight request observability, TTFT & tokens/sec
  • ▸ Export to JSON, JSONL, or Syslog for your SIEM
Explore observability →
0
PROVIDER KINDS
0
REGIONS AT GA
0
MAX FAILOVER ATTEMPTS
0
DB HITS ON THE GOVERNANCE HOT PATH
Watch it happen

One provider goes down. The session keeps running.

A kimi3 alias spread across four providers. DigitalOcean's stream dies mid-response. Trifrost continues the same SSE stream from the next healthy provider — and your user never knows.

Your app SSE stream tokens received: 0 Trifrost alias: kimi3 rr_health · failover ≤3 OpenRouter kimi3 · healthy DigitalOcean kimi3 · healthy Fireworks kimi3 · healthy Together kimi3 · healthy stream cut — zero content forwarded → marked unhealthy despite the 200 session uninterrupted — same SSE stream half-open probe → healthy → rejoining rotation
request response tokens healthy cooldown (30s) half-open probe

1 · Healthy rotation

The kimi3 alias round-robins across OpenRouter, DigitalOcean, Fireworks, and Together. Health is tracked in memory; unhealthy members sit out for a 30s cooldown.

2 · Provider fails mid-stream

DigitalOcean's stream ends with zero content forwarded. Trifrost marks it unhealthy despite the 200 and continues the same SSE stream from the next healthy provider. The client never sees the blip.

3 · Half-open recovery

After cooldown, a probe re-tests DigitalOcean. Healthy → it rejoins rotation. Failed attempts land in attempt_trail — nothing is silent.

How failover works →

Drop-in compatible

Change the base URL. Keep your code.

Trifrost speaks the OpenAI wire format. Point your existing SDK at Trifrost, swap your API key for a virtual key, and you're governed — budgets, allowlists, audit, and failover all apply immediately.

  • OpenAI & OpenRouter SDK compatible — /v1/chat/completions, /v1/responses, /v1/embeddings
  • Reasoning models — reasoning_content / reasoning / thinking passed through untouched
  • Cost in the response headers — X-TF-Cost-USD, token counts, model, provider
before → after
# before — direct to a provider
export OPENAI_BASE_URL=https://api.openai.com/v1
export OPENAI_API_KEY=sk-...

# after — governed by Trifrost
export OPENAI_BASE_URL=https://api.trifrost.tech/v1
export OPENAI_API_KEY=sk-bf-...  # your virtual key

# every call now: budgeted, allowlisted, audited,
# load-balanced, failover-protected, cost-attributed.
The Trifrost spend dashboard: cost over time, breakdown by provider and model, and budget consumption bars
Spend — by provider, model, department, user, or session
FinOps for AI

Know what a session costs. Not just a call.

Per-request cost is table stakes. Trifrost maps inference calls to sessions — so you can answer "what did that coding session cost?" and "which department is burning Opus?" in one query.

The FinOps story →

Shipping now

Universal gateway · alias-group load balancing + zero-content failover · governance hierarchy with per-model budgets · model allowlists · HMAC-signed audit log · OIDC/SSO with claim-driven teams · playground · session cost attribution · spend forecast + alerts · five regions.

Ready when you are

Run AI like infrastructure. Not like a science project.

Book a demo See pricing

Deployed in AU · APAC · EMEA · UK · DE. Need somewhere else? We'll stand it up.