Models, providers and routing
Automatic fallback
When a model or its provider goes down, an alias moves to its fallback and tells you why.
What it is
Give an alias a fallback model. When the target is disabled, its upstream fails a probe, or recent requests to it are all failing, requests are served by the fallback.
Why you want it
A provider outage should not stop every team that depends on one model. Fallback keeps work moving and tells you it happened.
Stable names, what each resolves to right now, its fallback, and 30 days of spend, requests and callers.
Actual Janus interface · Demo identities and synthetic usageHow it works
- Triggers: target unavailable, upstream unreachable, model down (3+ failures in 10 minutes), optional error rate of 10% or more
- One caller's failed request never reroutes everyone
- Responses carry X-Janus-Fallback-Reason and the request log records the serving model
- Both down returns 503 policy.managed_model_fallback_exhausted naming both faults
In the product
Screenshot above. Actual Janus interface with demo identities and synthetic usage.
See it on your own network.
The Community edition is free for up to 25 people. The 30-day Business trial unlocks every Business feature.