Omni Runtime
Checking runtime
Console Get API Key
OpenAI-compatible AI Gateway

One clean gateway for model access and token truth.

Omni gives agents and apps a single OpenAI-compatible entry, then turns every request into a readable audit receipt: routing, token source, estimated cost, latency, provider claim, and context changes.

Get an API Key View report demo
Unified accessUse one Base URL for OpenAI-compatible clients.
/v1
Token auditSeparate reported tokens, estimates, cost, and confidence.
reported
Provider transparencyShow requested model, resolved route, latency, and model claim status.
trace
Fast path first

Default requests stay lightweight: route, stream, measure, and report.

BYOK direction

Users should bring their own provider keys; sponsored models stay invite-gated.

Metadata privacy

Reports avoid storing raw prompts, responses, auth headers, and provider secrets.

Runtime receipts

Every successful route can become a readable operational receipt.

AI Gateway Runtime Transparency

Request Audit Receipt

One OpenAI-compatible entry for routing, token visibility, cost estimates, latency, and provider-claim review.

Omni Runtime records a metadata-only report for each OpenAI-compatible request: token source, cost availability, latency, provider claim, context delta, and review signals.

sample visible

Runtime Report

waiting metadata-only
loading
Report
-
Trace
-
Created
-

Reports are metadata-only by default. Raw prompts, responses, authorization headers, and provider secrets are not copied into the report store.

Token Usage
-
Waiting for report.
Estimated Cost
-
Waiting for report.
Latency
-
Waiting for report.
Provider / Model
-
Waiting for report.
Context Delta
-
Waiting for report.

Recent Reports

0 retained
No reports yetwaiting
Run one chat completion to populate evidence.

Token Breakdown

source unavailable
Segment Tokens Source Confidence
No report loaded.

Provider / Model Claim

unverified
Provider substrate
-
Provider name
-
Requested model
-
Resolved model
-
Profile
-
Verification
-

Context Delta

not measured
Mode
-
Before
-
After
-
Saved
-

Request Timeline

0 ms

Connect Flow

sample visible
1
Set Base URL
Use the Railway MVP endpoint as the OpenAI-compatible base URL.
2
Add Omni API key
Paste the key in this console or configure it in your client.
3
Run one request
The report will replace this public sample after the first routed request.

Quick Start

/health/readiness 
from openai import OpenAI

client = OpenAI(
    base_url="https://omni-gateway-api-production.up.railway.app/v1",
    api_key="YOUR_OMNI_KEY"
)

response = client.chat.completions.create(
    model="omni-runtime/smoke",
    messages=[{"role": "user", "content": "Hello Omni"}]
)

# Reports endpoint: /_omni/reports

Boundary

Codex routed: false
Privacy
metadata-only
Runtime mode
-
Provider call
false