alibaba/spring-ai-alibaba · critical · RuntimeException

All models failed after ${fallbackModels.size() + 1} attempt

Error message

All models failed after ${fallbackModels.size() + 1} attempts

What it means

ModelFallbackInterceptor tried the primary model (the intercepted handler) and every configured fallback ChatModel, and all of them threw. It exhausts the chain and rethrows a RuntimeException whose message reports the total attempt count (fallbackModels.size() + 1) with the last failure as the cause. The real root cause is in the suppressed cause chain of the last model's exception.

Source

Thrown at spring-ai-alibaba-agent-framework/src/main/java/com/alibaba/cloud/ai/graph/agent/interceptor/modelfallback/ModelFallbackInterceptor.java:100

		for (int i = 0; i < fallbackModels.size(); i++) {
			ChatModel fallbackModel = fallbackModels.get(i);
			try {
				log.info("Trying fallback model {} of {}", i + 1, fallbackModels.size());

				// Call the fallback model directly
				Prompt prompt = new Prompt(request.getMessages(), request.getOptions());
				var response = fallbackModel.call(prompt);

				return ModelResponse.of(response.getResult().getOutput());
			}
			catch (Exception e) {
				log.warn("Fallback model {} failed: {}", i + 1, e.getMessage());
				lastException = e;
			}
		}

		// All models failed
		throw new RuntimeException("All models failed after " + (fallbackModels.size() + 1) + " attempts", lastException);
	}

	@Override
	public String getName() {
		return "ModelFallback";
	}

	public static class Builder {
		private final List<ChatModel> fallbackModels = new ArrayList<>();

		public Builder addFallbackModel(ChatModel model) {
			this.fallbackModels.add(model);
			return this;
		}

		public Builder fallbackModels(List<ChatModel> models) {
			this.fallbackModels.addAll(models);
			return this;

View on GitHub (pinned to f82da0b50f)

Solutions

1. Inspect the cause chain (getCause()/lastException) of this RuntimeException to find the actual per-model failure and fix that root issue first.
2. Verify each fallback model's credentials, base URL, model name, and quota independently — test each ChatModel with a trivial 'ping' call at startup.
3. Add fallbacks from a different provider or deployment region so a single outage does not take down the whole chain.
4. Check the intercepted response text: if it contains 'Exception:' it is treated as failure — make sure the primary handler is not returning error text for benign reasons.

Example fix

// before: all fallbacks share one key/provider
ModelFallbackInterceptor.builder()
    .addFallbackModel(dashScopeMini)
    .addFallbackModel(dashScopeTurbo)
    .build();
// after: diversify providers
ModelFallbackInterceptor.builder()
    .addFallbackModel(openAiGpt4oMini)
    .addFallbackModel(deepseekChat)
    .build();

List<ChatModel> fallbacks = builderModels;
if (fallbacks.isEmpty()) throw new IllegalStateException("Configure at least one fallback model");
fallbacks.forEach(m -> {
    try { m.call(new Prompt("ping")); } catch (Exception e) { throw new IllegalStateException("Fallback model unhealthy: " + e.getMessage(), e); }
});

try {
    return agentCall();
} catch (RuntimeException e) {
    if (e.getMessage() != null && e.getMessage().startsWith("All models failed after")) {
        log.error("All models exhausted; root cause: {}", e.getCause(), e.getCause());
        return degradedResponse(); // serve cached/stub answer, alert ops
    }
    throw e;
}

When it happens

Trigger: Calling ModelFallbackInterceptor.interceptModel(request, handler) when handler.call(request) throws (or returns text containing 'Exception:') and every fallbackModel.call(prompt) also throws — e.g. all endpoints down, all API keys invalid/quota-exhausted, or a shared network outage.

Common situations: Provider-wide outages or rate limiting hitting primary and fallbacks from the same account; misconfigured base URLs or expired API keys shared across all configured models; fallback models pointing at the same unavailable cluster as the primary; regional network blocking of all provider endpoints.

Related errors

• Model call failed, maximum number of retries reached:${excep
• Model call failed, maximum number of retries reached.
• Model call failed, maximum number of retries reached. ${maxA
• Failed to subscribe
• Oauth2CallError

AI-assisted analysis of alibaba/spring-ai-alibaba@f82da0b50f (2026-09-09). Data as JSON: /api/errors/a881c619430ba410. Report an issue: GitHub.