Tasks

A task is one job your application asks a model to do, such as "summarize this document", "classify this ticket", or "extract these fields". You give each task a name, and Inference.net groups every request with that name together.

Tasks group requests by purpose, not by model or prompt. You can change the model or rewrite the prompt for a task, and its history stays in one place.

#Why use tasks

When your application has more than one AI feature, tasks let you:

• See requests, cost, and success rate for each feature separately.
• Compare the models each feature uses.
• Change the model or prompt for one feature without mixing its numbers with the others.

#Create a task

You create a task by sending a request with the x-inference-task-id header. You do not need to set it up first. The task appears in the dashboard when the first request with that header arrives.

Use a short, descriptive name in lowercase with hyphens, such as document-summary, ticket-classifier, or extract-product-data. Every request that uses the same name belongs to the same task.

#Set the task on each request

Use this when one client sends requests for more than one task.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: process.env.INFERENCE_API_KEY,
});

const response = await client.chat.completions.create(
  {
    model: "glm-5.2",
    messages: [{ role: "user", content: "Summarize this document..." }],
  },
  {
    headers: { "x-inference-task-id": "document-summary" },
  },
);

#Set the task for a whole client

Use this when a client only sends requests for one task. Put the header in the client's default headers, and every request from that client belongs to the task.

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.inference.net/v1",
  apiKey: process.env.INFERENCE_API_KEY,
  defaultHeaders: { "x-inference-task-id": "document-summary" },
});

#The default task

Requests without an x-inference-task-id header go into the default task. A header set to default does the same. These requests are still recorded, but they all share one group. Add task names to see each feature on its own.

#See your tasks

Open Usage in the dashboard and select Tasks. Each task shows its requests, cost, success rate, and the models it used for the selected time range.

Select a task to see its usage over time. To hide a task you no longer use, archive it from the task list. You can unarchive it later.