chroma-core/chroma · error · ValueError

No embeddings returned from the API

Error message

No embeddings returned from the API

What it means

After a successful embed_content call, GoogleGeminiEmbeddingFunction validates the response shape: response must have a non-empty 'embeddings' attribute. An HTTP-200 response with no embeddings violates the API contract, so the function refuses to return an empty/partial result. This is almost always SDK-version drift or a backend anomaly rather than a user-input problem.

Source

Thrown at chromadb/utils/embedding_functions/google_embedding_function.py:114

        from google.genai.types import EmbedContentConfig

        config = EmbedContentConfig(
            task_type=self.task_type,
            output_dimensionality=self.dimension,
        )

        try:
            response = self.client.models.embed_content(
                model=self.model_name,
                contents=input,
                config=config,
            )
        except Exception as e:
            raise ValueError(f"Failed to generate embeddings: {str(e)}") from e

        # Validate response structure
        if not hasattr(response, "embeddings") or not response.embeddings:
            raise ValueError("No embeddings returned from the API")

        embeddings_list = []
        for ce in response.embeddings:
            if not hasattr(ce, "values"):
                raise ValueError("Malformed embedding response: missing 'values'")
            embeddings_list.append(np.array(ce.values, dtype=np.float32))

        return cast(Embeddings, embeddings_list)

    @staticmethod
    def name() -> str:
        return "google_gemini"

    def default_space(self) -> Space:
        return "cosine"

    def supported_spaces(self) -> List[Space]:
        return ["cosine", "l2", "ip"]

View on GitHub (pinned to aecdd12c8a)

Solutions

1. Pin google-genai to a version known to work with your chromadb release (check chromadb's optional-dependency pins)
2. If it occurs in tests, make the mock return an object with embeddings=[...] where each item has .values
3. Retry once - transient backend anomalies can clear
4. Report with the raw response if it persists across pinned versions

try:
    vecs = ef(docs)
except ValueError as e:
    if "No embeddings returned" in str(e):
        # response-shape anomaly (SDK drift/backend glitch): capture context, retry once
        vecs = ef(docs)
    else:
        raise

When it happens

Trigger: A google-genai version whose response object lacks or empties the 'embeddings' field; a Gemini backend anomaly returning an empty embedding list; a mocked/stubbed client in tests that returns an object without .embeddings.

Common situations: Upgrading google-genai to a release with changed response types; test doubles that mimic the client incompletely; rare upstream incidents.

Related errors

• Malformed embedding response: missing 'values'
• The google-genai python package is not installed. Please ins
• The {self.api_key_env_var} environment variable must be set
• Failed to generate embeddings: {str(e)}
• Unknown error

AI-assisted analysis of chroma-core/chroma@aecdd12c8a (2026-08-16). Data as JSON: /api/errors/10f8b0d8abc2acd8. Report an issue: GitHub.