Need Help: "No tool output found for function call" Error After Function Call in Responses API

API
1

Hi all,

I’m working with the Responses API for my project and I’m encountering an issue after a function call.

Here’s how I’ve set up my tool definitions:

tools = [
    { "type": "web_search_preview" },
    {
        "type": "file_search",
        "vector_store_ids": ["store_id_1"],
    },
    {
        "type": "function",
        "name": "xxx",  
        "description": "xxxx",
        "strict": True,
        "parameters": {
            "type": "object",
            "properties": {},
            "required": [],
            "additionalProperties": False        
        }
    }
]

Here’s my API call logic:

api_arguments = {
    "model": self.model,
    "input": messages,
    "temperature": self.temperature
}

if tools:
    api_arguments["tools"] = tools

if previous_response_id:
    api_arguments["previous_response_id"] = previous_response_id

response = self.client.responses.create(**api_arguments)

After a function call, I am appending follow-up messages like this:

# Append the model's function call message
follow_up_messages = [output_item]

# Append the function call output
follow_up_messages.append({
    "type": "function_call_output",
    "call_id": output_item.call_id,
    "output": str(return_val)
})

logger.info(f"output_item: {output_item}")

# Request a follow-up response
follow_up_response = st.session_state.open_ai.get_llm_responses(
    follow_up_messages,
    nexi_tools, 
    None
)

The problem:
After the function call, when I try to input something new, I get this error:

*** Error occurred while question prompt: Error code: 400 - {'error': {'message': 'No tool output found for function call call_xxx.', 'type': 'invalid_request_error', 'param': 'input', 'code': None}}

It seems like the function call output isn’t being linked correctly, but I’m not sure what I’m missing.

Could someone please help me figure this out?

Thank you,
Binjan Iyer

2

I figure out the solution of the problem. I did not send back the result from function call. It does not have all the information. So, OpenAI Reponses API return error.

append only function result for follow-up response

# Append the function call output
follow_up_messages.append({
    "type": "function_call_output",
    "call_id": output_item.call_id,
    "output": str(return_val)
})

# Request a follow-up response
follow_up_response = st.session_state.open_ai.get_llm_responses(
    follow_up_messages,
    nexi_tools, 
    api_arguments["previous_response_id"]
)
3

I added what you suggested to my code and also managed the tool according to the official documentation. But this silly error:
*** Error occurred while question prompt: Error code: 400 – {‘error’: {‘message’: ‘No tool output found for function call call_xxx.’, ‘type’: ‘invalid_request_error’, ‘param’: ‘input’, ‘code’: None}}
still keeps happening. What’s frustrating is that it’s becoming more and more frequent.

4

The model may call tools repeatedly, so you have to chain the output of previous responses to the next one:

import OpenAI from "openai";
const openai = new OpenAI({ apiKey: process.env.OPENAI_API_KEY });

async function runConversation() {
  // 1️⃣ First request — send user/system messages
  let response = await openai.responses.create({
    model: "gpt-4.1",
    input: [
      { role: "system", content: "You can call tools to answer questions." },
      { role: "user", content: "Check my calendar and then message me the result." }
    ],
    tools: [
      { name: "check_calendar", type: "function", function: { parameters: {} } },
      { name: "send_message", type: "function", function: { parameters: { message: "string" } } }
    ],
    tool_choice: "auto"
  });

  // 2️⃣ Loop: resolve tool calls until no more
  while (true) {
    const calls = response.output.filter(o => o.type === "function_call");
    if (!calls.length) break;

    // Run all tools and build outputs
    const outputs = await Promise.all(
      calls.map(async (call: any) => {
        let result;
        if (call.name === "check_calendar") {
          result = { available: ["2025-08-15 10:00"] };
        } else if (call.name === "send_message") {
          result = { status: "sent" };
        }
        return {
          type: "function_call_output",
          call_id: call.call_id,
          output: JSON.stringify(result)
        };
      })
    );

    // 3️⃣ Follow-up request — chain with previous_response_id
    response = await openai.responses.create({
      model: "gpt-4.1",
      previous_response_id: response.id,
      input: outputs,
      tools: [
        { name: "check_calendar", type: "function", function: { parameters: {} } },
        { name: "send_message", type: "function", function: { parameters: { message: "string" } } }
      ],
      tool_choice: "auto"
    });
  }

  console.log("Final model output:", response.output_text);
}

runConversation();