Models & Auto-Selector

What you'll do

Every chat runs on a model, and you decide which one. You will open the model picker, either leave Auto selected or choose a specific model, and know what to do when a model you expect is not available.

Step-by-step

1. Open a chat and open the model picker from the chat input. It shows the model currently in use.
2. Check that AI has not been switched off for this chat — the control is in the picker.
3. Leave Auto selected, or choose a specific model from one of the groups.
4. Type your message and send it.

Auto, or a specific model

Auto is the default selection. Leave it on and Emerge.ai picks a model for each request, so you get a good fit without thinking about it.

Choose a specific model when you want every message in a chat answered by the same one. Both are choices you make in the picker — Auto is an option you select, not something that happens invisibly.

The groups in the picker

The picker organizes models into named groups — the outside providers by name, plus Emerge.ai's own. As of the last check, that means:

• Anthropic, Google, OpenAI and xAI — the outside providers, by name
• Emerge Private Hosted — models Emerge.ai hosts itself
• Emerge Fast & Free — the free model chat continues on at a usage limit

For the models in each group, see Model selector.

Emerge Private Hosted

Emerge.ai hosts these models itself, so your request stays with Emerge.ai instead of going to an outside provider.

That is also why the group matters for privacy. With Lockdown Mode on, chat runs only on Emerge.ai's private models — which set exactly is configuration Emerge.ai changes over time, so check the picker to see what you can use.

Emerge Fast & Free

The model in the Emerge Fast & Free group is Qwen 3.5 - 9b as of the last check, and on plans that include the group, reaching a usage limit does not stop your work: at your cap, chat keeps going on this model and Emerge.ai tells you it has switched.

See Plans, Tokens & Billing for how the daily and weekly limits are measured and what else you can do when you reach one.

When a model isn't available

Four different things produce this, and they look different in the picker:

• It isn't included in your plan. It still appears in the picker, marked unavailable with a note saying it is not on your current plan. Choose another model, or upgrade.
• It's temporarily unavailable. Choose another model, or try again later.
• An administrator has hidden it for the whole Cloud. Administrators can hide specific models from Cloud Settings → AI Models, whatever the plan allows. A model hidden this way disappears from the picker entirely for everyone in that Cloud, so it is missing rather than marked unavailable.
• Lockdown Mode is on. Only the models Emerge.ai allows under Lockdown Mode can be used, and Auto is unavailable — the option leaves the picker, and a chat set to Auto moves to a specific allowed model.

Downloadable models on mobile

On the mobile apps you can download a model to the device and run it there. Chat then works fully offline — on a plane, or anywhere without a connection — and there is no limit on how much you use a downloaded model.

Which models can be downloaded is shown in the app itself — that list is the live one, and it changes.

See Platform & Apps for what else the mobile apps add.

Local models on desktop (beta)

The Emerge.ai desktop app for Mac and Windows is in beta. On paid plans, it can chat with open models running on your own computer, through the free Ollama app or a model server you already use like LM Studio, with no cost per message.

• Pick per chat. Switch between cloud and local models in the model picker. Skills work with local models.
• Recommended for your computer. Emerge.ai checks your computer's memory and suggests the best models for it, with download sizes and speed estimates, and shows the license and terms before anything downloads.
• Doesn't count toward your limits. Local model replies are free and don't use your plan's AI allowance. Your chat history still saves to your Cloud like any other chat.
• Can be turned off per Cloud. Cloud Settings has a switch to allow or turn off local models for a Cloud.

See Platform & Apps for the desktop app itself.

Definitions

• Auto — the default model-picker option, which lets Emerge.ai choose a model for each request instead of you fixing one for the whole chat.
• Emerge Private Hosted — a picker group of models Emerge.ai hosts itself, offering capability comparable to the frontier models.
• Emerge Fast & Free — a picker group holding the free model that chat continues on once you reach a usage limit.
• Local model — a model that runs on your own device. In the mobile apps it is downloaded to the device, so chat works with no connection. In the desktop app (beta), on paid plans, it runs on your computer.
• Lockdown Mode — a setting that limits which models can be used and makes Auto unavailable.

Related topics

• Model selector — the models in each group, and when the list was last checked against the product
• Your first chat — opening the picker and sending a first message
• Plans, Tokens & Billing — usage limits, and the two ways forward at a cap
• Lockdown Mode — which models remain usable, and why Auto is not one of them
• Platform & Apps overview — web, mobile, and offline use
• Troubleshooting