Fallback Chains

Try multiple models in order. If the first is unavailable, automatically fall back to the next.

Using fallback_models

{
  "model": "my-preferred-model",
  "fallback_models": ["gemma-4-31B-it-Q8_0.gguf", "auto"],
  "messages": [...]
}Copy

OpenRouter-style models[] array

{
  "models": ["model-a", "model-b", "auto"],
  "messages": [...]
}Copy

LLMFinder tries each model in order, skipping any with no available hosters or open circuit breakers.