Skip to content
Book a demo

Every model call, on your terms.

One control plane for your AI connective tissue. Route, govern, observe, and ship to production faster, without rewiring your stack every time a provider changes.

Get started in 5 minRead the docs
drop-in replacement
import os
from openai import OpenAI
 
client = OpenAI(
    base_url="https://gw.to11.ai/v1",
    api_key=os.environ["OPENAI_API_KEY"],  # provider key, forwarded upstream
    default_headers={"x-to11-authorization": "Bearer " + os.environ["TO11_API_KEY"]},
)
# every request now routes, traces & guards.
res = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello from to11"}],
)
Two lines to route every request through the gateway

01Why to11

Move quickly, stay in control.

  • Go to production faster

    Swap models, add fallbacks, and roll out changes from config, not a redeploy.

  • Centrally control and govern your AI connective tissue

    One place to set policy, manage credentials, and watch cost across every team, app, and provider.

02What it can do

Everything a model call needs.

Every request through the gateway is routed, recorded, priced, and checked against policy, so reliability and governance stop being per-app work.

routing · chat-production
primary    claude-sonnet-4
fallback   gpt-5             on 5xx, timeout
backup     gemini-2.5-pro    duplicate slow requests
a/b        gpt-5-mini        30% of traffic
canary     claude-opus-4     5% of users in eu-west
guard      pii, policy       block before response
Illustrative routing rules

  • Reliability

    Routing, fallbacks, automatic backups, and canary releases keep calls answered when a provider blinks.

    • Uptime with backup requests
    • Fallbacks across providers
    • Routing rules per app & environment
    • Canary releases

  • Quality & experimentation

    Catch failures before users do, and test changes against real traffic.

    • Catch failures
    • A/B test across models
    • A/B test across users & geos

  • Observability & cost

    Record every trace. Monitor cost in real time. Know exactly what each call cost and why.

  • Safety & governance

    Guardrails block policy violations. A secure credential store keeps keys out of app code.

    • Guardrails for policy violations
    • Secure credential store
    • Workspace, project & environment isolation
    • Enterprise-grade RBAC

03Strengths

Fast where it counts. Flexible everywhere else.

  • Lightning fast

    Rust

    Written in Rust. The gateway adds latency you’ll have to look hard to find.

  • Easy to adopt

    Passthrough · managed credentials · routing

    Start with passthrough, move to managed credentials, then full routing. Start where you are, move when you’re ready.

  • Extremely flexible

    Cross-provider

    Cross-provider compatible, with a broad scope of configuration options. However you’ve wired your AI, the gateway fits.

04Coming soon

What’s next for the gateway.

  • Coming Soon

    Open source

    The gateway, in the open.

  • Coming Soon

    500+ providers with cost data

    Route to more providers, with pricing built in.

  • Coming Soon

    Sophisticated routing rules

    Richer conditions for how and where each request goes.

Point your traffic through to11.

One control plane for every model call. Route, govern, and observe without rewiring your stack.

Get started in 5 minBook a demo