vercel/ai · error · UnsupportedFunctionalityError

AI_UnsupportedFunctionalityError

AI_UnsupportedFunctionalityError

Error message

'tool choice type: ${_exhaustiveCheck}' functionality not supported.

What it means

prepareTools maps LanguageModelV4 toolChoice types ('auto', 'none', 'required', 'tool') to Bedrock toolConfig.toolChoice. The default branch is an exhaustiveness guard using UnsupportedFunctionalityError; it fires when a tool choice type has no Bedrock equivalent mapping. Usually means a new tool choice type from the core SDK not yet supported by the Bedrock provider, or a bad cast.

Source

Thrown at packages/amazon-bedrock/src/amazon-bedrock-prepare-tools.ts:214

  ) {
    const type = toolChoice.type;
    switch (type) {
      case 'auto':
        amazonBedrockToolChoice = { auto: {} };
        break;
      case 'required':
        amazonBedrockToolChoice = { any: {} };
        break;
      case 'none':
        amazonBedrockTools.length = 0;
        amazonBedrockToolChoice = undefined;
        break;
      case 'tool':
        amazonBedrockToolChoice = { tool: { name: toolChoice.toolName } };
        break;
      default: {
        const _exhaustiveCheck: never = type;
        throw new UnsupportedFunctionalityError({
          functionality: `tool choice type: ${_exhaustiveCheck}`,
        });
      }
    }
  }

  const toolConfig: AmazonBedrockToolConfiguration =
    amazonBedrockTools.length > 0
      ? { tools: amazonBedrockTools, toolChoice: amazonBedrockToolChoice }
      : {};

  return {
    toolConfig,
    additionalTools,
    betas,
    toolWarnings,
  };
}

View on GitHub (pinned to 69428b1f8b)

Solutions

1. Update @ai-sdk/amazon-bedrock (and `ai`) to matching latest versions
2. Use one of the supported toolChoice values: 'auto', 'none', 'required', or { type: 'tool', toolName: '...' }
3. Remove type assertions that let invalid toolChoice values through
4. If a new core tool-choice type is unsupported, file an issue on the bedrock provider

Example fix

// before
await generateText({ model, tools, toolChoice: { type: 'any' } as any });
// after
await generateText({ model, tools, toolChoice: 'required' });

const ok = ['auto','none','required'].includes(toolChoice as string) ||
  (typeof toolChoice === 'object' && toolChoice?.type === 'tool' && !!toolChoice.toolName);
if (!ok) throw new Error('Unsupported toolChoice for Bedrock');

function isSupportedToolChoice(tc: unknown): boolean {
  return tc === 'auto' || tc === 'none' || tc === 'required' ||
    (typeof tc === 'object' && tc !== null && (tc as any).type === 'tool');
}

try {
  await generateText({ model, tools, toolChoice });
} catch (e) {
  if (UnsupportedFunctionalityError.isInstance(e)) {
    // fall back to toolChoice: 'auto'
  }
}

When it happens

Trigger: Calling generateText/streamText with toolChoice set to a value the bedrock provider can't map — typically via an untyped/`as any` toolChoice object, or a newer core SDK tool-choice type with an older @ai-sdk/amazon-bedrock.

Common situations: Version skew between `ai` and `@ai-sdk/amazon-bedrock`; hand-built toolChoice objects with typos like 'any' or 'force'; programmatically generated toolChoice from config files.

Related errors

• AI_UnsupportedFunctionalityError
• Unsupported task type: ${taskType}
• URL-based images are not supported for Amazon Bedrock image
• Multiple system messages that are separated by user/assistan
• file parts with provider references

AI-assisted analysis of vercel/ai@69428b1f8b (2026-08-30). Data as JSON: /api/errors/35d8924e74cd0b77. Report an issue: GitHub.