Tutorial 13: Choosing Models for Your AI

This is part of the Hermes Agent official tutorial series. View all tutorials

Think of your AI assistant like a restaurant kitchen: you have a head chef who handles every dish that comes in, and a team of sous-chefs who take care of smaller prep tasks so the main chef can focus. In Hermes, you get to pick who your head chef is—and who your sous-chefs are—all from one simple dashboard.

Step 1: Open the Models Page

After launching your Hermes dashboard, click Models in the sidebar. You’ll see two sections:

• Model Settings (top panel) — this is where you assign models to different “slots”
• Usage Analytics (below) — a ranked list of every model you’ve used, with token counts, costs, and capability badges

Think of the top panel as your hiring board. The bottom is your payroll report.

Step 2: Set Your Main Model (The Head Chef)

The Main model is what your agent thinks with. Every user message, every tool-call loop, every streamed response flows through this model. It’s the brain of your operation.

Click Change on the Main model row. A picker dialog opens with two columns:

• Left column — only providers you’ve already set up (API keys added, OAuth’d, or custom endpoints defined). If a provider is missing, head to Keys and add its credential first.
• Right column — the curated list of agentic models Hermes recommends for that provider. This isn’t the raw dump of every model (which on OpenRouter includes 400+ models like TTS and image generators). Hermes filters it down to models that actually work well for agent tasks.

Type in the filter box to narrow by provider name, slug, or model ID. Pick a model, hit Switch, and Hermes writes it to your config file automatically.

Important: This applies to new sessions only. Any chat tab you already have open keeps running whatever model it started with. To hot-swap the current chat, use the /model slash command inside it.

Step 3: Understand Auxiliary Models (The Sous-Chefs)

Hermes uses a second kind of slot for auxiliary models — smaller side-jobs the agent offloads so your main model isn’t overloaded. These include:

• Context compression — shrinking long conversations to fit memory limits
• Vision — analyzing images
• Web-page summarization — condensing articles
• Approval scoring — deciding if a tool call is safe
• MCP tool routing — picking which tools to use
• Session-title generation — naming your chats
• Skill search — finding relevant skills

Each auxiliary slot has its own model and can be overridden independently. You don’t need to touch these if you’re just starting out—Hermes picks sensible defaults. But if you notice your main model is doing too much heavy lifting, you can offload tasks to cheaper, faster models here.

Step 4: Mid-Session Switches and Warnings

When you switch models inside an active session (using the Hermes TUI model picker, the hermes CLI, or /model on Telegram/Discord), Hermes estimates whether your next message will trigger preflight context compression against the new model’s window. If your session is already near or above that model’s compression threshold, Hermes will warn you before you switch—so you don’t accidentally lose context.

Step 5: The Fastest Path — Nous Portal

If you want 300+ models under one subscription, try the Nous Portal. On a fresh install, run:

hermes setup --portal

This logs you in and sets Nous as your provider in one command. Check what’s wired up with:

hermes portal info

Portal subscribers also get 10% off token-billed providers. It’s the easiest way to get started without juggling multiple API keys.

A Quick Note on Config Files

On a brand-new install, your config has model: "" (an empty string meaning “not configured yet”). The first time you run hermes setup or hermes model, that key upgrades to a mapping with provider, default, base_url, and api_mode sub-keys. If you ever see an empty string in config.yaml, just run hermes model (or click Change in the dashboard) and Hermes will write the proper format for you.

Summary

Choosing models in Hermes is like staffing your kitchen: pick one head chef for the main thinking, and let smaller models handle prep work. The dashboard makes it visual, the CLI makes it fast, and the Nous Portal makes it one-command simple. Start with the main model, leave auxiliaries at their defaults, and tweak as you learn what your workflows need.

Next up: Tutorial 14 — “Managing API Keys and Providers” — where we’ll walk through connecting your first provider and keeping your credentials safe.