issue: Workspace > Models: import of single model not working without response #4597

New Issue

2025-11-11T15:58:07-06:00

GiteaMirror commented

2025-11-11 15:58:07 -06:00

Originally created by @MaxStroh on GitHub (Mar 28, 2025).

Check Existing Issues

I have searched the existing issues and discussions.
I am using the latest version of Open WebUI.

Installation Method

Other

Open WebUI Version

v0.5.20

Ollama Version (if applicable)

No response

Operating System

Kubernetes / Ubuntu

Browser (if applicable)

Chrome 134.0.6998.166, Firefox 136.0.4

Confirmation

I have read and followed all instructions in README.md.
I am using the latest version of both Open WebUI and Ollama.
I have included the browser console logs.
I have included the Docker container logs.
I have listed steps to reproduce the bug in detail.

Expected Behavior

When exporting a single model from workspace > models I would expect to be able to import this single model again on another instance (e.g. staging / prod environment)
When the import of a model json file fails I would like to get a dedicated exception

Actual Behavior

Exporting a single model via Workspace > Models > "More" / "Three dots"-Button on a single model > Export downloads a (seemingly) valid model JSON file
When importing the model JSON file on another instance via the Workspace > Models > "Import Models"-Button on the right of the page, I get some logs in the browser console, logging the model included in the JSON file, but then nothing happens
No error message, no error log in the browser console - simply nothing is happening
Even in the running pod there are no logs about the failed import

Steps to Reproduce

Go to 'Workspace' > 'Models'
If no custom model exists, create a new demo model (settings are irrelevant, breaks with any settings)
Save the newly created model
On any existing model on 'Workspace' > 'Models' click on the "three dots"-Button (Tooltip labeled with "More")
Click on "Export" and save the json file
On the page 'Workspace' > 'Models' click on the 'Import models' button in the bottom right
Choose the downloaded json file from earlier and confirm
Inspect the browser console if you want

Logs & Screenshots

Export single model:

Import JSON file:

Browser Logs:

Additional Information

When looking into the code and comparing to the json file downloaded when exporting ALL models via the button on the bottom right on the 'Workspace' > 'Models' page I think the models json file of the single model is missing an encapsulation into an "info" field.
In the all models json file, there are all model details and attributes (including the custom models created via 'Workspace' > 'Models') encapsuled in an "info" field like this:

[
  // ... other models here...
  {
    "id": "demo-model",
    "name": "Demo Model",
    "object": "model",
    "created": 1743178699,
    "owned_by": "openai",
    "info": {
      "id": "demo-model",
      "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87",
      "base_model_id": "azure-gpt-4o-mini",
      "name": "Demo Model",
      "params": {
        "system": "Answer politely"
      },
      "meta": {
        "profile_image_url": "/static/favicon.png",
        "description": "Demo description",
        "capabilities": {
          "vision": true,
          "usage": false,
          "citations": true
        },
        "suggestion_prompts": null,
        "tags": []
      },
      "access_control": null,
      "is_active": true,
      "updated_at": 1743178699,
      "created_at": 1743178699
    },
    "preset": true,
    "actions": []
  }
]

The single model export file is missing this attribute "info":

[
  {
    "id": "demo-model",
    "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87",
    "base_model_id": "azure-gpt-4o-mini",
    "name": "Demo Model",
    "params": {
      "system": "Answer politely"
    },
    "meta": {
      "profile_image_url": "/static/favicon.png",
      "description": "Demo description",
      "capabilities": {
        "vision": true,
        "usage": false,
        "citations": true
      },
      "suggestion_prompts": null,
      "tags": []
    },
    "access_control": null,
    "is_active": true,
    "updated_at": 1743178699,
    "created_at": 1743178699,
    "user": {
      "id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87",
      "name": "Test User",
      "email": "test@example.com",
      "role": "admin",
      "profile_image_url": "/user.png"
    }
  }
]

Also, I found the following code snippet in the code, which was a hint, why the single exported model file is not working:
https://github.com/open-webui/open-webui/blob/b03fc97e287f31ad07bda896143959bc4413f7d2/src/lib/components/workspace/Models.svelte#L420-L434
So my suggestion is, that inside the for loop, it is looked for a field named "info", if it is missing - simply nothing is done, the content of the model config is skipped and nothing happens
So I thought "what if I just wrap the whole content of the single exported json file inside a field named "info" - et voilà: it worked! I was able to import the following single-model-file via the 'Workspace' > 'Models' > 'Import models' button without any errors:

[
  {
    "info": {
      "id": "demo-model",
      "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87",
      "base_model_id": "azure-gpt-4o-mini",
      "name": "Demo Model",
      "params": {
        "system": "Answer politely"
      },
      "meta": {
        "profile_image_url": "/static/favicon.png",
        "description": "Demo description",
        "capabilities": {
          "vision": true,
          "usage": false,
          "citations": true
        },
        "suggestion_prompts": null,
        "tags": []
      },
      "access_control": null,
      "is_active": true,
      "updated_at": 1743178699,
      "created_at": 1743178699,
      "user": {
        "id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87",
        "name": "Test User",
        "email": "test@example.com",
        "role": "admin",
        "profile_image_url": "/user.png"
      }
    }
  }
]

So my suggestion would be: When exporting a single model, wrap the config into an "info" field and export this as json
Or: Modify the if clause in the code to not only check, if the file includes an "info" field, but instead check otherwise, whether the file includes a model configuration

Short disclaimer:

We have OpenWebUI deployed and running on a kubernetes cluster inside a pod. So unfortunately I did not test this locally (but depending on the code, this should also appear locally on Docker)
The logs of the pod did log exactly nothing during this time, so I did not include these
We do not use Ollama, but also this should not be a problem of the base model imo

Exported single model json file (not working):
demo-model-1743179018399.json

Export all models json file (working, but includes all models):
models-export-1743179408340.json

Modified (and able to import) single model json file:
modified-demo-model.json

Originally created by @MaxStroh on GitHub (Mar 28, 2025). ### Check Existing Issues - [x] I have searched the existing issues and discussions. - [x] I am using the latest version of Open WebUI. ### Installation Method Other ### Open WebUI Version v0.5.20 ### Ollama Version (if applicable) _No response_ ### Operating System Kubernetes / Ubuntu ### Browser (if applicable) Chrome 134.0.6998.166, Firefox 136.0.4 ### Confirmation - [x] I have read and followed all instructions in `README.md`. - [x] I am using the latest version of **both** Open WebUI and Ollama. - [x] I have included the browser console logs. - [x] I have included the Docker container logs. - [x] I have listed steps to reproduce the bug in detail. ### Expected Behavior * When exporting a single model from workspace > models I would expect to be able to import this single model again on another instance (e.g. staging / prod environment) * When the import of a model json file fails I would like to get a dedicated exception ### Actual Behavior * Exporting a single model via Workspace > Models > "More" / "Three dots"-Button on a single model > Export downloads a (seemingly) valid model JSON file * When importing the model JSON file on another instance via the Workspace > Models > "Import Models"-Button on the right of the page, I get some logs in the browser console, logging the model included in the JSON file, but then nothing happens * No error message, no error log in the browser console - simply nothing is happening * Even in the running pod there are no logs about the failed import ### Steps to Reproduce 1. Go to 'Workspace' > 'Models' 2. If no custom model exists, create a new demo model (settings are irrelevant, breaks with any settings) 3. Save the newly created model 4. On any existing model on 'Workspace' > 'Models' click on the "three dots"-Button (Tooltip labeled with "More") 5. Click on "Export" and save the json file 6. On the page 'Workspace' > 'Models' click on the 'Import models' button in the bottom right 7. Choose the downloaded json file from earlier and confirm 8. Inspect the browser console if you want ### Logs & Screenshots Export single model: <img width="968" alt="Image" src="https://github.com/user-attachments/assets/6df69eff-c66a-46c4-b6bd-82e2261e34de" /> Import JSON file: <img width="1702" alt="Image" src="https://github.com/user-attachments/assets/7d968708-c4a1-47f7-9f8d-b61a29653aef" /> Browser Logs: <img width="1641" alt="Image" src="https://github.com/user-attachments/assets/5627800f-8df3-4d30-903c-7d071fe07d9b" /> ### Additional Information * When looking into the code and comparing to the json file downloaded when exporting ALL models via the button on the bottom right on the 'Workspace' > 'Models' page I think the models json file of the single model is missing an encapsulation into an "info" field. * In the all models json file, there are all model details and attributes (including the custom models created via 'Workspace' > 'Models') encapsuled in an "info" field like this: ``` json [ // ... other models here... { "id": "demo-model", "name": "Demo Model", "object": "model", "created": 1743178699, "owned_by": "openai", "info": { "id": "demo-model", "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87", "base_model_id": "azure-gpt-4o-mini", "name": "Demo Model", "params": { "system": "Answer politely" }, "meta": { "profile_image_url": "/static/favicon.png", "description": "Demo description", "capabilities": { "vision": true, "usage": false, "citations": true }, "suggestion_prompts": null, "tags": [] }, "access_control": null, "is_active": true, "updated_at": 1743178699, "created_at": 1743178699 }, "preset": true, "actions": [] } ] ``` * The single model export file is missing this attribute "info": ``` json [ { "id": "demo-model", "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87", "base_model_id": "azure-gpt-4o-mini", "name": "Demo Model", "params": { "system": "Answer politely" }, "meta": { "profile_image_url": "/static/favicon.png", "description": "Demo description", "capabilities": { "vision": true, "usage": false, "citations": true }, "suggestion_prompts": null, "tags": [] }, "access_control": null, "is_active": true, "updated_at": 1743178699, "created_at": 1743178699, "user": { "id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87", "name": "Test User", "email": "test@example.com", "role": "admin", "profile_image_url": "/user.png" } } ] ``` * Also, I found the following code snippet in the code, which was a hint, why the single exported model file is not working: https://github.com/open-webui/open-webui/blob/b03fc97e287f31ad07bda896143959bc4413f7d2/src/lib/components/workspace/Models.svelte#L420-L434 * So my suggestion is, that inside the for loop, it is looked for a field named "info", if it is missing - simply nothing is done, the content of the model config is skipped and nothing happens * So I thought "what if I just wrap the whole content of the single exported json file inside a field named "info" - et voilà: it worked! I was able to import the following single-model-file via the 'Workspace' > 'Models' > 'Import models' button without any errors: ```json [ { "info": { "id": "demo-model", "user_id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87", "base_model_id": "azure-gpt-4o-mini", "name": "Demo Model", "params": { "system": "Answer politely" }, "meta": { "profile_image_url": "/static/favicon.png", "description": "Demo description", "capabilities": { "vision": true, "usage": false, "citations": true }, "suggestion_prompts": null, "tags": [] }, "access_control": null, "is_active": true, "updated_at": 1743178699, "created_at": 1743178699, "user": { "id": "73ae4aa4-b7df-4bcc-abb7-7b86d6480b87", "name": "Test User", "email": "test@example.com", "role": "admin", "profile_image_url": "/user.png" } } } ] ``` * So my suggestion would be: When exporting a single model, wrap the config into an "info" field and export this as json * Or: Modify the if clause in the code to not only check, if the file includes an "info" field, but instead check otherwise, whether the file includes a model configuration --- Short disclaimer: * We have OpenWebUI deployed and running on a kubernetes cluster inside a pod. So unfortunately I did not test this locally (but depending on the code, this should also appear locally on Docker) * The logs of the pod did log exactly nothing during this time, so I did not include these * We do not use Ollama, but also this should not be a problem of the base model imo --- Exported single model json file (not working): [demo-model-1743179018399.json](https://github.com/user-attachments/files/19509168/demo-model-1743179018399.json) Export all models json file (working, but includes all models): [models-export-1743179408340.json](https://github.com/user-attachments/files/19509187/models-export-1743179408340.json) Modified (and able to import) single model json file: [modified-demo-model.json](https://github.com/user-attachments/files/19509403/modified-demo-model.json)

GiteaMirror added the bug label 2025-11-11 15:58:07 -06:00

GiteaMirror closed this issue

2025-11-11 15:58:07 -06:00

GiteaMirror commented

2025-11-11 15:58:08 -06:00