Integration API

Developer Reference

Consume WyzBench evaluation results programmatically. Use the Recommendation and WyzRouter Config endpoints to integrate model-selection intelligence into Wyzard or any downstream system.

🔑 Authentication

All endpoints require an X-API-Key header. Your API key is scoped to your organisation — it cannot access other orgs' data. Keys are long-lived; rotate via your team admin.

Base URL
https://wyzbench.34.56.55.184.nip.io
Header format
X-API-Key: wb_your_key_here

⚡ Rate Limits

TierReq / minReq / day
Dev key10500
Production6010,000

429 responses include a Retry-After header (seconds). Only POST /v1/evaluations is rate-limited; GET endpoints are unlimited.

# Respect Retry-After on 429
HTTP/1.1 429 Too Many Requests
Retry-After: 30

# Retry after the header value:
sleep 30 && curl -H "X-API-Key: ..." \
  https://wyzbench.34.56.55.184.nip.io/v1/evaluations

📦 Versioning

All endpoints are under /v1/. Breaking changes increment the version. The current stable version is v1.

🔀 WyzRouter Integration Pattern

The recommended integration pattern is:

  1. Submit a WyzBench evaluation for your task type once (or after a supersession alert)
  2. Fetch the router config — GET /v1/evaluations/{job_id}/router-config
  3. Store the config in your WyzRouter or routing layer
  4. At inference time, WyzRouter selects primary_model with automatic fallback to fallback_model
  5. Re-run evaluations on a schedule or when a supersession nudge is received
Configure provider credentials →
Filter:8 endpoints
Need help?

Questions about the integration API — contact your WyzBench account team.

Provider SettingsRun Evaluation →