Skip to main content

​
Overview

orchid01 supports OpenAI-compatible tool calling. Define tools using JSON Schema, pass them in the request, and the model returns structured tool calls when it needs to use them. This works with any tool definition you’d use with OpenAI — including MCP tools, LangChain tools, and custom functions.

​
Basic example

tools = [{
    "type": "function",
    "function": {
        "name":        "get_sec_filing",
        "description": "Retrieve an SEC filing by ticker and form type",
        "parameters": {
            "type": "object",
            "properties": {
                "ticker":    {"type": "string", "description": "Company ticker symbol"},
                "form_type": {"type": "string", "enum": ["10-K", "10-Q", "8-K", "13-F"]},
            },
            "required": ["ticker", "form_type"],
        },
    },
}]

response = client.chat.completions.create(
    model="orchid01",
    messages=[{"role": "user", "content": "Get Apple's latest 10-K"}],
    tools=tools,
    temperature=0.1,
)

tool_call = response.choices[0].message.tool_calls[0]
print(tool_call.function.name)       # get_sec_filing
print(tool_call.function.arguments)  # {"ticker": "AAPL", "form_type": "10-K"}

​
Full agentic loop

import json

messages = [{"role": "user", "content": "Analyse Apple's latest 10-K for covenant risks"}]

response = client.chat.completions.create(
    model="orchid01",
    messages=messages,
    tools=tools,
)

if response.choices[0].message.tool_calls:
    tool_call = response.choices[0].message.tool_calls[0]
    args      = json.loads(tool_call.function.arguments)
    
    # Execute your tool
    result = get_sec_filing(**args)
    
    # Send result back
    messages.append(response.choices[0].message)
    messages.append({
        "role":         "tool",
        "tool_call_id": tool_call.id,
        "content":      json.dumps(result),
    })
    
    # Get final response
    final = client.chat.completions.create(
        model="orchid01",
        messages=messages,
        tools=tools,
        extra_body={"orchid": {"dehallucinate": True}},
    )
    print(final.choices[0].message.content)

​
MCP tools

orchid01 works with MCP (Model Context Protocol) tools. Your MCP server runs client-side — translate its tool definitions into JSON Schema format and pass them in the request as shown above. Popular finance MCP servers that work with orchid01:
  • EdgarTools — 13 tools for SEC EDGAR filings, financial statements, insider trading, live filings feed. No API key required.
  • Bloomberg — connects to your existing Bloomberg Terminal via blpapi.
  • Alpha Vantage — real-time and historical market data.

​
LangChain agents

from langchain_openai  import ChatOpenAI
from langchain.agents  import AgentExecutor, create_openai_tools_agent
from langchain.tools   import tool

llm = ChatOpenAI(
    model="orchid01",
    base_url="https://llm.orchid.ac/v1",
    api_key="orchid-your-key-here",
    temperature=0.1,
)

@tool
def get_covenant_data(company: str) -> str:
    """Get covenant data for a company from internal database"""
    # your implementation
    ...

agent = create_openai_tools_agent(llm, [get_covenant_data], prompt)
executor = AgentExecutor(agent=agent, tools=[get_covenant_data])
executor.invoke({"input": "Check covenant headroom for Citadel"})

​
Tips

  • Use low temperature (0.0–0.2) for tool calling — reduces hallucinated parameter values
  • Write detailed tool descriptions — the model decides when to call a tool based on the description
  • Use tool_choice="required" to force the model to always call a tool