support for Decimal in google gemini genai

[oegedijk]

• This is actually a bug report.
• I am not getting good LLM Results
• I have tried asking for help in the community on discord or discussions and have not received a response.
• I have tried searching the documentation and have not found an answer.

What Model are you using?

• gpt-3.5-turbo
• gpt-4-turbo
• gpt-4
• Other (gemini-2.5-flash)

Describe the bug
When using a Decimal in a schema instructor complains that genai does not support Optional types, even though the genai client seems to work with Decimals just fine?

To Reproduce

from decimal import Decimal

import instructor
from google import genai
from google.genai import types
from pydantic import BaseModel

genai_client = genai.Client()


class Receipt(BaseModel):
    total: Decimal


response = genai_client.models.generate_content(
    model="gemini-2.5-flash",
    contents=[
        "Extract out the total and line items from the invoice",
        "the total on the receipt is 100.34",
    ],
    config=types.GenerateContentConfig(
        response_mime_type="application/json",
        response_schema=Receipt,
    ),
)

print(response.text)

for mode in [instructor.Mode.GENAI_STRUCTURED_OUTPUTS, instructor.Mode.GENAI_TOOLS]:
    try:
        instructor_client = instructor.from_genai(genai_client, mode=mode)

        response = instructor_client.chat.completions.create(
            model="gemini-2.5-flash",
            response_model=Receipt,
            messages=[
                {
                    "role": "user",
                    "content": [
                        "Extract out the total and line items from the invoice",
                        "the total on the receipt is 100.34",
                    ],
                },
            ],
            )

        print(response)
    except Exception as e:
        print(e)
        print(f"Mode {mode} failed")

results in:

{"total":100.34}
Gemini does not support Union types (except Optional). Please change your function schema
Mode Mode.GENAI_STRUCTURED_OUTPUTS failed
Gemini does not support Union types (except Optional). Please change your function schema
Mode Mode.GENAI_TOOLS failed

Expected behavior
A clear and concise description of what you expected to happen.

Screenshots
If applicable, add screenshots to help explain your problem.

[oegedijk]

guess the issue is that Decimal renders to json schema

{'properties': {'total': {'anyOf': [{'type': 'number'}, {'type': 'string'}], 'title': 'Total'}}, 'required': ['total'], 'title': 'Receipt', 'type': 'object'}

which then gets tripped up by

def verify_no_unions(obj: dict[str, Any]) -> bool:
    """
    Verify that the object does not contain any Union types (except Optional).
    Optional[T] is allowed as it becomes Union[T, None].
    """
    for prop_value in obj["properties"].values():
        if "anyOf" in prop_value:
            # Check if this is an Optional type (Union with None/null)
            any_of_list = prop_value["anyOf"]
            if not isinstance(any_of_list, list) or len(any_of_list) != 2:
                return False

            # Check if one of the types is null (representing None in Optional[T])
            has_null = any(
                isinstance(item, dict) and item.get("type") == "null"
                for item in any_of_list
            )

            if not has_null:
                # This is a true Union type, not Optional - reject it
                return False

        if "properties" in prop_value and not verify_no_unions(prop_value):
            return False

    return True