Skip to main content
Physical fiber routes entering a data-center exchange.

Model access

One gateway for your registered model catalog

Use the models currently registered to your Routeon workspace through one compatible entry point.

OpenAI
DeepSeek

Developer path

Integrate once. Keep routing and evidence in the platform.

Keep familiar calls while retaining request evidence.

  1. 01

    Prepare credentials

    Limit access to the current project and required models.

    ROUTEON_API_KEY
  2. 02

    Choose a route

    Start with auto, then pin a model by cost or latency.

    model: auto
  3. 03

    Verify the result

    Follow the trace ID through request, latency, and error evidence.

    trace_id
ROUTEON / QUICKSTART
1import OpenAI from 'openai';
2
3const client = new OpenAI({
4 apiKey: process.env.ROUTEON_API_KEY,
5 baseURL: 'https://api.routeon.dev/v1',
6});
7
8await client.chat.completions.create({
9 model: 'auto',
10 messages: [{ role: 'user', content: 'Inspect this trace' }],
11});
Request evidence200 OK · route policy: cost-first · trace: trc_8f2a9c4e

Enterprise governance

Keep access, cost, and accountability on one evidence chain.

Turn access, cost, and accountability into operable controls.

Workspaces

Isolate resources and accountability by team and environment.

Roles

Constrain console and operations actions by role.

Key scope

Give every key an explicit scope, state, and lifecycle.

Budget boundary

Connect usage, budgets, and bills when data is available.

Audit evidence

Record critical operations and request context.

Risk state

Expose unknown, pending, and blocking states explicitly.

Capability map

One control plane from integration through operations.

Connect access, routing, governance, and commercial boundaries.

accessCONNECTED

Use one compatible entry point for a governed model catalog.

  • Model capability and catalog management stay centralized
  • Clear protocol boundaries reduce switching cost
View model catalog →

Route evidence

Every selection has an explainable path.

Keep policy and outcomes traceable from request to metering.

Structure illustration; use the console for live data
  1. 01Request

    Receive model, key, scope, and call context.

  2. 02Policy

    Apply visibility, permission, region, and routing rules.

  3. 03Candidates

    Filter providers and models that meet current constraints.

  4. 04Score

    Evaluate cost, latency, and availability from available evidence.

  5. 05Selection

    Record the final channel, model, and decision result.

  6. 06Failover

    Choose a bounded fallback path after upstream failure.

  7. 07Meter

    Record request, token, and extended-usage evidence.

  8. 08Return

    Return the unified response and trace context to the caller.

Commercial and operating boundary

Operate with explicit rules, not vague promises.

Account capabilities follow current configuration and permissions.

FAQ

Questions before you connect

Answers grounded in current product boundaries.

Does Routeon replace provider accounts?

No. Upstream access follows the channels, credentials, and permissions configured for your Routeon environment.

Which models are available?

Availability follows the current reviewed model catalog and upstream state. Use the model catalog for account-specific evidence.

Does the homepage show pricing?

Only when pricing can be verified for the current account and configuration. This public page does not invent price or discount claims.

Can Routeon choose a model automatically?

Automatic selection is bounded by the routing policies, visible models, and evidence configured for the request context.

Where can I inspect a routing decision?

Use request identifiers, logs, observability, and route evidence available to your account.

How do I start integrating?

Create and scope an API key, review the documentation, then validate a request and its trace evidence.

Next step

Choose a verifiable starting point.

Create a key, read the docs, or check system status.