assafelovic/gpt-researcher · error · ValueError

Cost must be an integer or float

Error message

Cost must be an integer or float

What it means

Raised by GPTResearcher.add_costs when the cost argument is not a float or int (e.g. a string like '0.02' or None). The method accumulates research_costs and per-step step_costs, so it requires a real number to keep the arithmetic valid. It is a simple type precondition check.

Source

Thrown at gpt_researcher/agent.py:785

        Args:
            verbose: Whether to enable verbose output.
        """
        self.verbose = verbose

    def add_costs(self, cost: float) -> None:
        """Add to the accumulated API costs.

        The cost is attributed to the current step set via ``_current_step``.

        Args:
            cost: Cost amount to add in USD.

        Raises:
            ValueError: If cost is not a number.
        """
        if not isinstance(cost, (float, int)):
            raise ValueError("Cost must be an integer or float")
        self.research_costs += cost
        step = self._current_step
        self.step_costs[step] = self.step_costs.get(step, 0.0) + cost
        if self.log_handler:
            self._log_event("research", step="cost_update", details={
                "cost": cost,
                "total_cost": self.research_costs,
                "step_name": step,
            })

View on GitHub (pinned to 6f998577d5)

Solutions

1. Convert before calling: add_costs(float(cost)) or add_costs(Decimal->float)
2. If the value may be missing, default it: add_costs(cost or 0.0)
3. Wrap the call in try/except ValueError if cost provenance is untrusted

Example fix

// before
agent.add_costs(response["cost"])  # "0.02" string
// after
agent.add_costs(float(response["cost"]))

cost = float(cost) if cost is not None else 0.0

def is_numeric_cost(c) -> bool:
    return isinstance(c, (int, float)) and not isinstance(c, bool)

try:
    agent.add_costs(cost)
except ValueError:
    agent.add_costs(float(cost))

When it happens

Trigger: Calling agent.add_costs('0.5'), add_costs(None), or passing a Decimal/numpy type is fine only for int/float—strings and None raise. Common when cost is parsed from a JSON/config value that arrived as a string.

Common situations: Reading cost from an LLM response or env var as a string; computing costs from a dict like response.get('cost') that returns None; passing Decimal from a money library.

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

• Invalid type for path. Expected str, bytes, os.PathLike, or
• Embedding provider not found.
• Invalid retriever(s) found: {', '.join(invalid_retrievers)}.
• Set SMART_LLM or FAST_LLM = '<llm_provider>:<llm_model>' Eg
• Invalid reasoning effort: {reasoning_effort_str}. Valid opti

AI-assisted analysis of assafelovic/gpt-researcher@6f998577d5 (2026-08-28). Data as JSON: /api/errors/7b47088051d808fe. Report an issue: GitHub.