AISERVA and AIKuari
Manage AI Models
Model management in AISERVA lets you see every AI model installed on your connected computers, choose which chat model and which image model answer, install extra models, and remove ones you no longer need to free disk space. You do it from two places: the Models tab, which shows every server at once, and the Server Details window of a single server, which also installs new models. It is used by whoever looks after the AI server, usually the owner or an office administrator.
Before You Start
- At least one server is Connected. A server that is offline shows "Offline or unreachable right now - it reappears here once it reconnects."
- If you have no servers yet, the Models tab says No Connected Servers Yet and points you to Servers. See Connect Your AI Server.
- Enough free disk space and memory for the model you want. Each installable model shows what it needs.
How Models Work in AISERVA
Each connected computer has its own set of models, in two groups.
| Group | Used By | Hardware |
|---|---|---|
| Chat | Ask AI, ChatBots, InstaClaim receipt reading and any app that reads or writes text | Most modern computers; bigger models need more memory |
| Image | Dezigner and xDresser; Imaginator and Passport also need the extended image engine | Apple Silicon or an NVIDIA GPU; some models are marked NVIDIA Only |
Only one chat model and one image model are active on a computer at a time. The active one answers every request from every app. The others stay on disk, ready to switch to.
Model States
| State | What It Means | Buttons You See |
|---|---|---|
| Active (shown as "In use" on Models) | This model is answering right now | Turn Off on Models, Stop Using in Server Details |
| Installed | On disk and ready to switch to | Use, Remove |
| Not Downloaded | "Not downloaded - the transfer did not finish" | Remove on Models; Download Again or Remove in Server Details |
| Support Component | "Support component - it helps other models run"; cannot be turned on by itself | Remove |
Each model also shows how much space it uses on disk, and Server Details shows the total for each group, for example "Using 9.6 GB on this computer."
See All Your Models
- Go to the Models tab. The page reads "Every AI model on your connected computers. Turn one on, stop using it, or remove it to free disk space."
- Each connected server is listed with its Chat and Image groups.
- If a group is empty, the page says "No chat model installed" or "No image model installed" and points you to Servers.
Switch the Active Model
- On Models, find the model you want under the right server and group.
- Press Use. It becomes Active and the previous model steps aside. The page confirms "Turned on."
- The next request from any app is answered by the new model.
Stop Using a Model
Stopping a model frees memory and, for image models, the graphics card, without deleting anything.
- Press Turn Off next to the active model on Models, or Stop Using in Server Details.
- Read the confirmation. For chat it warns that "Chat and assistants will be unavailable" until you turn a model on. For images it warns that "Image generation pauses and the GPU is freed" until you turn a model on.
- Press Stop Using to confirm.
To bring it back, press Use on any ready model in that group.
Install a New Model
- Go to Servers and press Details on the server card.
- Under Chat Models or Image Generation Models, press the + Add button under that group, for example + Add a Chat Model.
- Read each card before installing:
| Card Detail | What It Tells You |
|---|---|
| Download size | How much disk space the model takes |
| Memory badge | The memory (for image models, graphics memory) it needs to run |
| NVIDIA Only | Runs only on an NVIDIA graphics card |
| Recommended for This Computer | The best fit the portal found for this computer's hardware |
| Estimated time per image | For image models, a rough time per picture on this computer |
- Press Install. The card shows the download progress.
- When it finishes, press Use to make it active.
Remove a Model
- On Models, press the bin icon (Remove) next to the model, or press Remove in Server Details.
- Confirm "Remove ... from this computer to free disk space?" by pressing Remove.
Server Service Controls
In Server Details you can also control the AI service on that computer:
| Button | What It Does |
|---|---|
| Start | Starts the AI service if it is stopped |
| Stop | Stops the AI service; apps cannot use this server until you start it |
| Restart | Stops and starts the service; a good first fix if answers stop |
| Rescan | Refreshes the list of installed models and the status |
| Update | Updates the connector software, shown only when an update is useful |
Frequently Asked Questions
Why Does the First Image Take So Long?
The first image after the image engine starts can take a few minutes to warm up. Later images are much faster.
Where Are Models Stored?
On the disk of your own computer. The Models page shows how much space each one uses.
Can Two Servers Run Different Models?
Yes. Each connected server has its own models and its own active choice.
Why Can I Not Install an Image Model?
Image generation needs Apple Silicon or an NVIDIA GPU. On other computers Server Details says "Image generation needs Apple Silicon or an NVIDIA GPU - not available on this computer."
Does Stopping a Model Delete It?
No. Turn Off and Stop Using only unload it. Only Remove deletes it from disk.