vercel/ai · error · InvalidArgumentError

targetLanguage is required for translation model '${this.mod

Error message

targetLanguage is required for translation model '${this.modelId}'.

What it means

The speech-translation model's doStream requires options.targetLanguage because a translation model must know which language to translate into; Gemini Live translation cannot infer it. When targetLanguage is null or undefined it throws an InvalidArgumentError naming the model. This is a pre-flight argument validation thrown before any network or WebSocket connection is made.

Source

Thrown at packages/google/src/speech-translation/google-speech-translation-model.ts:121

  }

  get provider(): string {
    return this.config.provider;
  }

  constructor(
    modelId: GoogleSpeechTranslationModelId,
    config: GoogleSpeechTranslationModelConfig,
  ) {
    this.modelId = modelId;
    this.config = config;
  }

  async doStream(
    options: SpeechTranslationModelV4StreamOptions,
  ): Promise<Awaited<ReturnType<SpeechTranslationModelV4['doStream']>>> {
    if (options.targetLanguage == null) {
      throw new InvalidArgumentError({
        argument: 'targetLanguage',
        message: `targetLanguage is required for translation model '${this.modelId}'.`,
      });
    }

    const currentDate = this.config._internal?.currentDate?.() ?? new Date();
    const googleOptions = await parseProviderOptions({
      provider: 'google',
      providerOptions: options.providerOptions,
      schema: googleSpeechTranslationModelOptions,
    });
    const warnings: SharedV4Warning[] = [];

    validateGoogleSpeechTranslationInputAudioFormat(options.inputAudioFormat);

    if (options.sourceLanguage != null) {
      warnings.push({
        type: 'unsupported',

View on GitHub (pinned to 69428b1f8b)

Solutions

1. Pass targetLanguage (e.g. 'es', 'fr-FR') in the stream options.
2. If you only need a transcript, use a transcription model instead of a translation model so no target language is needed.
3. Validate options.targetLanguage before calling when model choice is dynamic.

Example fix

// before
await model.doStream({ audio: pcmStream });
// after
await model.doStream({ audio: pcmStream, targetLanguage: 'es' });

if (options.targetLanguage == null) {
  throw new Error('targetLanguage must be provided for translation models');
}
await model.doStream({ ...options, targetLanguage: options.targetLanguage });

function hasTargetLanguage(
  o: SpeechTranslationModelV4StreamOptions,
): o is SpeechTranslationModelV4StreamOptions & { targetLanguage: string } {
  return o.targetLanguage != null;
}

When it happens

Trigger: Calling experimental_streamTranscribe (or the speech-translation doStream path) on a translation model (e.g. a gemini live translation model id) without passing targetLanguage in the stream options.

Common situations: Reusing transcription code (which has no targetLanguage) against a translation model; switching model IDs from a transcribe model to a translation model without updating options; TypeScript not flagging it because targetLanguage is optional in the shared options type.

Understand the failure class

Background: "missing required argument" and "the following required arguments were not provided": what required-argument errors mean and how to fix them — this error's family across 20 libraries.

Related errors

• The Gemini Live translation API only supports 16kHz 16-bit P
• Google batch input files must not exceed 2 GB.
• Google Generative AI API key is required for streaming trans
• maxEmbeddingsPerCall must be greater than 0
• maxInputBytesPerCall must be greater than 0

AI-assisted analysis of vercel/ai@69428b1f8b (2026-08-30). Data as JSON: /api/errors/dfc33dc3148129da. Report an issue: GitHub.