AgentNexus added provider-capacity classification and one-click fallback model switching so transient model availability problems show retry and model-switch actions instead of looking like runtime deployment failures.
Status
One-click fallback added
Category
Runtime
Date
May 12, 2026
Read time
2 min read
Impact
Operators can tell the difference between a temporary upstream model capacity issue and an AgentNexus deployment or configuration problem.
What changed
Agent chat now classifies provider capacity and overload errors as retryable provider availability events. The API returns a structured capacity code, and the dashboard explains that the issue is transient and not a deployment failure.
The chat panel now gives users concrete next steps: retry shortly or switch directly to a suggested fallback model when the selected model is temporarily unavailable.
Why it matters
Model providers can occasionally run out of serving capacity. Clear classification keeps teams from debugging healthy deployments when the right action is to retry or select a fallback model.
What to do
- Retry the chat request after a short wait when a provider-capacity message appears.
- Use the suggested fallback buttons to move the agent to an available model without leaving the chat surface.
- Treat runtime lifecycle errors separately from provider capacity messages.