bug: OpenAI O1 model requires max_completion_tokens parameter and custom model.json not loading

[imtuyethan]

Jan version

0.5.13

Describe the Bug

Two related issues:

1. When using OpenAI O1 model settings in Jan, an error occurs requiring max_completion_tokens as a setting. This appears to affect all O1 models including preview and mini variants.

2. Custom model.json files placed in AppData\Roaming\Jan\data\models\ folder are not being read by Jan. The app attempts to lock the folders but doesn't recognize the custom model.json files.

Steps to Reproduce

1. Configure Jan to use OpenAI O1 model
2. Attempt to run the model -> Receive error about max_completion_tokens requirement
3. Create custom model.json file with O1 settings
4. Place file in AppData\Roaming\Jan\data\models\
5. Restart Jan -> Custom model.json not detected

Screenshots / Logs

https://discord.com/channels/1107178041848909847/1327701677813399705

What is your OS?

• MacOS
• Windows
• Linux

[louis-menlo]

Scoped in #3786

[louis-menlo]

Issue 2: Won't be fixed. The GUI should support adding a model.

[louis-menlo]

Update:

This error wont be displayed with the latest fix, since engine request transform template already filtered out unsupported parameters