Spike Alerts

Status Healthy
Window Last 7d
No active spikes — last 7d healthy.
Cost (30d)
Tokens (30d)
Calls (30d)
Loading billing period...
Total Calls
Total Token Spend
Cost / Request
No usage yet this period.
↺ Refresh

Live Routing Metrics

Connected
Total Requests
0
Cost Saved
$0.00
Savings vs Baseline
0%
Period Cost
$0.00
SAVINGS vs BASELINE 0%
Actual: $0.000000 Baseline: $0.000000
Live Event Feed (last 8)
Waiting for routing events...
1
Configure
2
Run
3
Results

Step 1 — Choose a Cost Constraint

Select a routing preset. Each generates the routing rules for your workflow.

Models in workflow
gpt-4o-mini
$0.75/1K
total/1K tokens
gpt-4o
$12.50/1K
total/1K tokens
Routing to gpt-4o-mini by default saves up to 94%
Loading presets...

Step 2 — Run a Prompt

The routing engine evaluates your prompt against the constraint and picks the best agent.

Your Routing Rules

Test Prompt

Step 3 — Routing Decision

Here's what ConductorOS decided and why.

saved vs. baseline
Prompt tested
Prompt Complexity
/ 10
Routing Reason

OpenAI-Compatible API

Drop-in replacement for OpenAI's chat completions API. Set baseURL to https://vibecoderz.polsia.app/conductor

// Example: OpenAI SDK integration
const client = new OpenAI({
  baseURL: 'https://vibecoderz.polsia.app/conductor',
  apiKey: 'conductor-key', // unused in MVP
});

const response = await client.chat.completions.create({
  workflow: 'my-workflow',  // optional — uses first loaded workflow if omitted
  messages: [{ role: 'user', content: 'Hello, world!' }],
  tags: ['greeting'],
  // ... standard OpenAI params
});

// Response includes extra routing metadata
// response.routing.agent        → "cheap-fast"
// response.routing.reason       → "default (cheapest capable) → cheap-fast"
// response.routing.complexity   → 2
// response.routing.latency_ms   → 1423

Routing History

Full audit trail of every routing decision — searchable, filterable, paginated.

Loading history...

Routing Analytics

Total Requests
0
Success Rate
Cost Saved vs Baseline
Total Tokens
0
SAVINGS vs BASELINE (always use most expensive model) 0%
Actual: $0.000000 Baseline: $0.000000

Recent Events

No events recorded yet.

Response Latency (30 days)

per model

Average and p95 latency in milliseconds across the last 30 days of routing decisions.

No routing events in the last 30 days.
Daily Routing Trend (30 days)

One row per day for the last 30 days — values reconcile with the per-model breakdown below.

Calls Cost Estimated Savings
Weekly Aggregates (Last 8 weeks)

One row per week for the last 8 weeks — calls, cost, estimated savings, spike alert count, and p95 latency. Values reconcile with the per-model breakdown above.

Calls Cost Estimated Savings Spike alerts p95 latency
Recent Workflow Checkpoints (Last 20)

One row per durable checkpoint — the last successful model picked at each step. Click a row with a linked decision to open the per-rationale inspector drawer.

Loading checkpoints...

Routing Breakdown (30 days)

Per-model performance over the last 30 days of routing decisions. Click a column header to sort.

routing savings (30 days)
vs. always-premium baseline (claude-3-opus) · actual $0.000000 of $0.000000 · 0 models
No routing events in the last 30 days.

Spend per Team (30 days)

One row per team — set a team_id on the API key to isolate this key's usage and cost into its own slice. Rows labeled (unassigned) include traffic from keys without a team_id.

Loading...

API Keys

Create named API keys to authenticate against the ConductorOS API.

Loading keys...

Spend per API Key (30 days)

Top 10 keys by cost — shows which integrations are driving spend. Each row carries a 30-day sparkline.

Loading...