Back to Help Center

Models

Models are the engines AI workers use to reason, write, and decide how to use tools. You choose models from chat, AI worker settings, triggers, inbound channels, and other execution surfaces. Choose by looking beyond the name and checking indicators such as cost, instruction following, autonomy, and conversation quality.

Overview

Understand how to choose models for AI workers and automations, including indicators and tradeoffs.

Basics

Reading the model menu

Open the model name in chat or settings to choose the model that fits the task. Each card shows cost, instruction following, autonomy, and conversation quality in one row. More ¥ marks mean higher cost; more dots mean the model is stronger on that trait.

What the indicators mean

Cost shows credit usage weight, instruction following shows how well the model follows detailed instructions, autonomy shows how well it decomposes and advances work, and conversation quality shows how naturally it can respond. The info icon stays to the left of Cost; hover it on desktop or tap it on mobile to see the indicator descriptions.

Basic vs Advanced

Start with Zemu Basic for everyday chat, support replies, short summaries, and light research. Use Zemu Advanced for long documents, multi-condition decisions, complex workflows, and important work where accuracy matters more.

After changing models

Changing the model alone does not guarantee stable workflow quality. Also review worker instructions, assigned knowledge, tools, skills, and expected output format. After changing models, compare output quality, speed, and cost on the same input before relying on it in operations.

Checklist

  1. Start with Zemu Basic and test a representative request.
  2. Switch to Zemu Advanced when the task needs complex reasoning or long-context work.
  3. After changing the model, re-test representative success paths in chat, triggers, and inbound channels.