What causes "params.values must be an array of strings" in vercel/ai?

Pass values as string[], e.g. ['first text', 'second text'].

params.values must be an array of strings

Error message

params.values must be an array of strings

What it means

The embedding model's doEmbed validates params.values is an array before sending to the performance client. A non-array values argument throws this error — a defensive guard against malformed embed() input.

Source

Thrown at packages/baseten/src/baseten-provider.ts:267

    if (!options.performanceClient) {
      return model;
    }

    // Opted in to the native client. It appends /v1 itself, so hand it the bare
    // /sync form.
    const performanceClient = new options.performanceClient(
      customURL.replace('/sync/v1', '/sync'),
      loadApiKey({
        apiKey: options.apiKey,
        environmentVariableName: 'BASETEN_API_KEY',
        description: 'Baseten API key',
      }),
    );

    model.doEmbed = async params => {
      if (!params.values || !Array.isArray(params.values)) {
        throw new Error('params.values must be an array of strings');
      }

      const response = await performanceClient.embed(
        params.values,
        // model_id is for Model APIs; dedicated deployments ignore it.
        modelId ?? 'embeddings',
      );

      return {
        embeddings: response.data.map(item => item.embedding),
        // The native client types its response as `any`; only report usage when
        // a token count is actually present rather than `{ tokens: undefined }`.
        usage:
          typeof response.usage?.total_tokens === 'number'
            ? { tokens: response.usage.total_tokens }
            : undefined,
        response: { headers: {}, body: response },
        warnings: [],

View on GitHub (pinned to 69428b1f8b)

Solutions

Pass values as string[], e.g. ['first text', 'second text'].
Coerce single strings to an array before calling embed().
Ensure the values argument is defined and not misnamed.

Example fix

// before
await embed({ model, values: 'hello world' });
// after
await embed({ model, values: ['hello world'] });

Defensive patterns

Strategy: validation

Validate before calling

function assertStringArray(values) {
  if (!Array.isArray(values) || values.some(v => typeof v !== 'string')) {
    throw new TypeError('values must be string[]');
  }
}
assertStringArray(values);

Type guard

function isStringArray(v: unknown): v is string[] {
  return Array.isArray(v) && v.every(x => typeof x === 'string');
}

Try / catch

try {
  await embed({ model, values });
} catch (e) {
  if (e instanceof Error && e.message === 'params.values must be an array of strings') {
    // coerce: values = Array.isArray(values) ? values : [String(values)];
  }
}

Prevention

Always pass an array to embed(), even for a single string.
Type the values parameter as string[] end-to-end.
Validate dynamic input shape before calling embed().

When it happens

Trigger: Calling embed({ model: baseten.textEmbeddingModel(...), values: 'string' }) or values: undefined / a non-array value.

Common situations: Passing a single string instead of an array; passing a variable that is undefined due to upstream bug; dynamic values built incorrectly.

Understand the failure class

Background: Invalid argument type errors: "must be of type string", "expected X, got Y", and ERR_INVALID_ARG_TYPE explained — this error's family across 15 libraries.

Related errors

AI-assisted analysis of vercel/ai@69428b1f8b (2026-08-30). Data as JSON: /api/errors/7bc97da1185bc94f. Report an issue: GitHub.