I use OpenAI 4o-mini for most of my AI automations and agents, but somethimes it fails because of huge context, I want 4o-mini to be the main model, but if it gets a context error, I want it to fallback to Gemini 2.0Flash.
Kinda have an error route, but for the AI model node
This would be great!! Time to time in rare cases, the model APIs go down. Even if you put an aggressive retry on the node, it could still fail. It would be great if we could have a built in way to have a fallback model.
This would make our workflows much more robust and reliable.
Deep seek for example is so cheap and great to use. But the API is so unreliable and fails quite often.
(fallbacks can be achieved with the “continue using error output” but that makes things more messy having extra nodes)
