Issue: Cost showing up as 0 despite having tokens

[ryanlohyr]

Issue you'd like to raise.

Issue

The cost for my runs is always showing as 0.

Details

According to the LangSmith docs:

For LangSmith to accurately derive costs for an LLM run, you need to provide token counts.

However, I’m using OpenAI models (which are listed on the models page: link), so I expected the costs to be calculated automatically.

Question

What am I doing wrong? Why is the cost showing as 0?

Logs & Traces

Here is a log screenshot:

And here’s a trace of a single run (contains multiple LLM calls):

Suggestion:

No response

[angus-langchain]

Hi @ryanlohyr,

Can you provide an MRE

additionally, can you please provide what sdk version you are using

[ryanlohyr]

Hey @angus-langchain, yeap for sure

Here is my MRE

import 'dotenv/config';
import OpenAI from 'openai';
import { wrapOpenAI } from "langsmith/wrappers"; // traces openai calls
import { traceable } from "langsmith/traceable";

// Initialize OpenAI client
const openai = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
});

// Wrap OpenAI client with LangSmith
const wrappedOpenAI = wrapOpenAI(openai);

// Create a traceable function for the question-answering flow
const answerQuestion = traceable(
  async (question) => {
    console.log(`Question: ${question}`);
    
    const response = await wrappedOpenAI.responses.create({
      model: 'gpt-4o',
      instructions: 'You are a helpful assistant that answers questions clearly and concisely.',
      input: question,
    });
    
    return response;
  },
  { name: "answer_question" }
);

// Main function to run the example
async function main() {
  const question = process.argv[2] || "What is the capital of France?";
  
  try {
    const response = await answerQuestion(question);
    console.log(response.output_text);
  } catch (error) {
    console.error('Error:', error.message);
  }
}

main();

Here is the sdk versions im using

The cost output is still 0

[ryanlohyr]

hey updates on this? @angus-langchain

[jacoblee93]

Hi @ryanlohyr sorry for the delay - will look at this first thing tomorrow.

[angus-langchain]

@ryanlohyr Confirmed I can repro. This is related to the Responses API which we are working on supporting better