chroma-core/chroma · error · ValueError

The model name is required.

Error message

The model name is required.

What it means

GoogleGeminiEmbeddingFunction.build_from_config requires the persisted config dict to contain 'model_name'; it is the only required key (task_type, dimension, etc. all default). It returns None-safe gets for everything else, so a config dict lacking model_name cannot identify which Gemini model to instantiate and the builder refuses.

Source

Thrown at chromadb/utils/embedding_functions/google_embedding_function.py:145

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

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

    @staticmethod
    def build_from_config(config: Dict[str, Any]) -> "EmbeddingFunction[Documents]":
        model_name = config.get("model_name")
        task_type = config.get("task_type")
        dimension = config.get("dimension")
        api_key_env_var = config.get("api_key_env_var", "GEMINI_API_KEY")
        vertexai = config.get("vertexai")
        project = config.get("project")
        location = config.get("location")

        if model_name is None:
            raise ValueError("The model name is required.")

        return GoogleGeminiEmbeddingFunction(
            model_name=model_name,
            task_type=task_type,
            dimension=dimension,
            api_key_env_var=api_key_env_var,
            vertexai=vertexai,
            project=project,
            location=location,
        )

    def get_config(self) -> Dict[str, Any]:
        config: Dict[str, Any] = {
            "model_name": self.model_name,
            "api_key_env_var": self.api_key_env_var,
            "vertexai": self.vertexai,
            "project": self.project,
            "location": self.location,

View on GitHub (pinned to aecdd12c8a)

Solutions

1. Include 'model_name' in the config, e.g. {'model_name': 'gemini-embedding-001', ...}
2. Emit configs from a live instance via ef.get_config() instead of writing them by hand - it always includes model_name
3. Validate the dict against the 'google_gemini' schema (GoogleGeminiEmbeddingFunction.validate_config) before building

Example fix

# before
from chromadb.utils.embedding_functions import GoogleGeminiEmbeddingFunction
ef = GoogleGeminiEmbeddingFunction.build_from_config({"task_type": "RETRIEVAL_DOCUMENT"})  # ValueError

# after
ef = GoogleGeminiEmbeddingFunction.build_from_config({
    "model_name": "gemini-embedding-001",
    "task_type": "RETRIEVAL_DOCUMENT",
})

config = {"model_name": "gemini-embedding-001", "task_type": "RETRIEVAL_DOCUMENT"}
assert config.get("model_name"), "model_name is required in google_gemini config"
ef = GoogleGeminiEmbeddingFunction.build_from_config(config)

from typing import Any, TypeGuard

def has_model_name(cfg: Any) -> TypeGuard[dict]:
    return isinstance(cfg, dict) and isinstance(cfg.get("model_name"), str) and bool(cfg["model_name"])

When it happens

Trigger: Calling GoogleGeminiEmbeddingFunction.build_from_config({}) or any dict without a 'model_name' key; hand-written or migrated config dicts where the key was dropped or renamed (e.g. 'model' instead of 'model_name'); corrupted persisted collection metadata being replayed.

Common situations: Manually crafting the config payload passed to get_embedding_function; migrating configs between chromadb versions; tests that build partial configs; editing stored JSON configs by hand.

Related errors

• Invalid hash algorithm specified: {alg}
• The google-genai python package is not installed. Please ins
• Vertex AI and API key are mutually exclusive in the client i
• The {self.api_key_env_var} environment variable must be set
• Input documents cannot be empty

AI-assisted analysis of chroma-core/chroma@aecdd12c8a (2026-08-16). Data as JSON: /api/errors/b8b97a890c80cbd6. Report an issue: GitHub.