# Keep agents running through provider failures with automatic fallback models

> Agents and teams can now be configured with fallback models that activate automatically when the primary model fails, whether from rate limits, outages, context window overflows, or other retryable…

- Published: 2026-04-02
- Author: Agno Team
- Category: Changelog
- Canonical: https://agno-com-nine.vercel.app/articles/keep-agents-running-through-provider-failures-with-automatic-fallback-models
- Markdown: https://agno-com-nine.vercel.app/articles/keep-agents-running-through-provider-failures-with-automatic-fallback-models.md

Agents and teams can now be configured with fallback models that activate automatically when the primary model fails, whether from rate limits, outages, context window overflows, or other retryable errors. Fallbacks are tried in order after the primary model’s retry loop is fully exhausted, and each fallback model runs its own independent retry cycle before the next one is attempted. Both simple lists and error-specific routing are supported, giving teams full control over how failures are handled.

Pass `fallback_models` to any Agent or Team. If the primary model fails after exhausting its retries, each fallback is tried in order until one succeeds.

If `gpt-4o` fails after exhausting its own retries, Claude is tried automatically.

[Model strings](https://docs.agno.com/models/model-as-string) work too:

See [Fallback Models docs](https://docs.agno.com/models/fallback-models) for more.