An error occurs in the Viz tool's chart interpretation when using OpenAI's gpt-5.x.

Thank you for creating such a great package :)

When using the gemini-3-flash-preview:cloud model in ollama, chart interpretation worked well.

However, when using OpenAI's gpt-5.x model, an error occurs.

I would appreciate it if you could fix this.

< code >
from pathlib import Path

from shiny import App, ui
from querychat import QueryChat
from querychat.data import titanic
from chatlas import ChatAzureOpenAICompletions

llm = ChatAzureOpenAICompletions(
api_key="",
endpoint="",
deployment_id="gpt-5.5",
api_version="2025-04-01-preview",
)

qc = QueryChat(
titanic,
"titanic",
client=llm,
tools=("query", "visualize"),
prompt_template=Path("prompt.md"),
)

app_ui = ui.page_fluid(qc.ui())

def server(input, output, session):
qc.server()

app = App(app_ui, server)

< Error >
('Error in Chat('querychat_CSCM_SALES_PSI_PV-chat'): Error code: 400 - {'error': {'message': "An assistant message with 'tool_calls' must be followed by tool messages responding to each 'tool_call_id'. The following tool_call_ids did not have response messages: call_RB81n3Dp1VpIsCXZ3bGfHpKp", 'type': 'invalid_request_error', 'param': 'messages.[7].role', 'code': None}}', False, False)

Start by reproducing the provided Python setup through QueryChat with tools=("query", "visualize") and qc.server(), using ChatAzureOpenAICompletions with the gpt-5.x deployment. Trace the chart interpretation path and its tool-call message handling based on the reported missing tool response. Done means gpt-5.x completes chart interpretation without the 400 error, while the working Ollama model remains functional.