ScrapeGraphAI/Scrapegraph-ai · error · ValueError

The model provided is not suppo

Error message

The model provided
                             is not supported. Supported models are:
                             {", ".join(supported_models)}.

What it means

GenerateAnswerFromImageNode.execute_async only supports OpenAI vision models; it reads node_config['config']['llm']['model'] (taking the part after the last '/'), and if it is not one of gpt-4o/gpt-4o-mini/gpt-4-turbo/gpt-4 it raises this ValueError listing supported models.

Source

Thrown at scrapegraphai/nodes/generate_answer_from_image_node.py:85

            )

    async def execute_async(self, state: dict) -> dict:
        """
        Processes images from the state, generates answers,
        consolidates the results, and updates the state asynchronously.
        """
        self.logger.info(f"--- Executing {self.node_name} Node ---")

        images = state.get("screenshots", [])
        analyses = []

        supported_models = ("gpt-4o", "gpt-4o-mini", "gpt-4-turbo", "gpt-4")

        if (
            self.node_config["config"]["llm"]["model"].split("/")[-1]
            not in supported_models
        ):
            raise ValueError(
                f"""The model provided
                             is not supported. Supported models are:
                             {", ".join(supported_models)}."""
            )

        api_key = self.node_config.get("config", {}).get("llm", {}).get("api_key", "")

        async with aiohttp.ClientSession() as session:
            tasks = [
                self.process_image(
                    session,
                    api_key,
                    image_data,
                    state.get("user_prompt", "Extract information from the image"),
                )
                for image_data in images
            ]

View on GitHub (pinned to 532dfffbf6)

Solutions

1. Set config.llm.model to a supported vision model, e.g. 'gpt-4o' or 'gpt-4o-mini'
2. Upgrade scrapegraphai — the supported list may have been extended
3. Ensure the model string's final '/'-segment matches the exact model name

Example fix

# before
llm_config = {'llm': {'model': 'openai/gpt-3.5-turbo', 'api_key': key}}
# after
llm_config = {'llm': {'model': 'openai/gpt-4o', 'api_key': key}}

SUPPORTED = ('gpt-4o', 'gpt-4o-mini', 'gpt-4-turbo', 'gpt-4')
model = llm_config['model'].split('/')[-1]
assert model in SUPPORTED, f'use one of {SUPPORTED} for image nodes'

def is_vision_model(model: str) -> bool:
    return model.split('/')[-1] in ('gpt-4o', 'gpt-4o-mini', 'gpt-4-turbo', 'gpt-4')

try:
    await node.execute_async(state)
except ValueError as e:
    if 'not supported' in str(e):
        llm_config['model'] = 'openai/gpt-4o'
    else:
        raise

When it happens

Trigger: Configuring the node with a non-vision model (e.g. 'gpt-3.5-turbo', 'claude-...', or an Azure deployment name); passing a provider-prefixed string whose suffix is unsupported.

Common situations: Reusing a general LLM config for an image graph; Azure deployment names that don't match the allowlist; model allowlist is hardcoded so newer models (gpt-4.1 etc.) also fail on older versions.

Related errors

• Adjacent state keys found without an operator between them.
• Invalid operator usage.
• Invalid operator placement: operators cannot be adjacent.
• Missing or unbalanced parentheses in expression.
• No state keys matched the expression.

AI-assisted analysis of ScrapeGraphAI/Scrapegraph-ai@532dfffbf6 (2026-08-28). Data as JSON: /api/errors/31e4a1a7213a51aa. Report an issue: GitHub.