n8n-io/n8n · error · NodeOperationError

The model "${modelName}" does not support the Sampling Tempe

Error message

The model "${modelName}" does not support the Sampling Temperature, Top K, or Top P options. Remove them from Options and try again.

What it means

A backstop error handler for the sampling-parameter deprecation. Some Anthropic models (and gateways whose capabilities the allow-list cannot infer) reject temperature/top_p/top_k with a 400 'deprecated'/'not supported' message. The handler matches /(temperature|top_p|top_k).*(deprecated|not supported)/i and converts the generic Bad Request into a message that tells the user to remove those sampling options.

Source

Thrown at packages/@n8n/nodes-langchain/nodes/llms/LMChatAnthropic/LmChatAnthropic.node.ts:620

						throw new NodeOperationError(
							this.getNode(),
							`The model "${modelName}" was not found at ${baseURL}. If you're using an AI gateway, select a model that your gateway supports.`,
							{ itemIndex },
						);
					}
				}
			: undefined;

		// Backstop for the sampling-parameter deprecation that modelSupportsSamplingParams
		// allow-lists against: catches the same 400 for gateway traffic (whose capabilities we
		// can't inspect from the model name) and for any Claude model the allow-list doesn't yet
		// account for, turning a generic "Bad request" into an actionable message.
		const deprecatedSamplingParamErrorHandler = (error: unknown) => {
			const message = error instanceof Error ? error.message : String(error);
			const isDeprecatedSamplingParamError =
				/(temperature|top_p|top_k).*(deprecated|not supported)/i.test(message);
			if (isDeprecatedSamplingParamError) {
				throw new NodeOperationError(
					this.getNode(),
					`The model "${modelName}" does not support the Sampling Temperature, Top K, or Top P options. Remove them from Options and try again.`,
					{ itemIndex },
				);
			}
		};

		const failedAttemptHandler = (error: unknown) => {
			gatewayErrorHandler?.(error);
			deprecatedSamplingParamErrorHandler(error);
		};

		const chatAnthropicParams: ChatAnthropicInput = {
			anthropicApiKey: credentials.apiKey,
			model: modelName,
			anthropicApiUrl: baseURL,
			maxTokens: options.maxTokensToSample,
			callbacks: [

View on GitHub (pinned to 5ac6606e81)

Solutions

1. Clear Temperature, Top K, and Top P in the node options for the offending model.
2. If you need control, switch to a model that still supports those parameters.
3. Upgrade the node typeVersion so modelSupportsSamplingParams allow-lists correctly where possible.

Example fix

// before
options = { temperature: 0.7, topP: 0.9 }
// after
options = {} // sampling params removed for unsupported model

// If using a model known to drop sampling params, clear them before request
const NO_SAMPLING = ['claude-opus-4-7', ...];
if (NO_SAMPLING.some(prefix => modelName.startsWith(prefix))) {
  delete options.temperature;
  delete options.topP;
  delete options.topK;
}

const isDeprecatedSamplingError = (e: unknown): boolean =>
  /(temperature|top_p|top_k).*(deprecated|not supported)/i.test(
    e instanceof Error ? e.message : String(e),
  );

try {
  await model.invoke(...);
} catch (e) {
  if (isDeprecatedSamplingError(e)) {
    // strip sampling options and retry once
  }
  throw e;
}

When it happens

Trigger: A Claude model that dropped legacy sampling params (or a gateway enforcing the new behaviour) is used while the user has Temperature, Top K, or Top P set in node options. The upstream 400 matches the regex and the backstop rewrites it.

Common situations: Anthropic deprecating sampling on newer Claude models; gateway strict mode; reusing an older workflow with sampling options against a newer model; expression-driven options always populating temperature.

Related errors

• No model selected. Please choose a model.
• Manual thinking mode is not supported on "${modelName}". Use
• VectorStore "${this.name}" requires a description — set it v
• The model "${modelName}" was not found at ${baseURL}. If you
• The value in the "Extra Body" field is not valid JSON

AI-assisted analysis of n8n-io/n8n@5ac6606e81 (2026-08-12). Data as JSON: /api/errors/82fefc406fdf13f7. Report an issue: GitHub.