Why does it take so long for Jan to update new models?

[imtuyethan]

How does Jan work when it comes to model releases? For example, can I use qwen 2.5 or GPT o-1 with it already? Why does it take so long to see new models in Jan?

[imtuyethan]

We're working on a big update to how Jan handles models. This should make it way easier for us to add new models in the future. Check out the details here: #3633

In the meantime, if you're itching to try new models:

1. For on-device models

You can import directly from Hugging Face.

2. For remote models (for example: GPT-o1), follow the steps here:

a) Create a new file called gpt-o1.json in your Jan models folder.
b) Paste this content into the file (adjust as needed for other models):

{
  "sources": [
    {
      "url": "https://openai.com"
    }
  ],
  "id": "o1-preview",
  "object": "model",
  "name": "o1-preview",
  "version": "1.1",
  "description": "o1-preview is a new flagship model with fast speed and high quality",
  "format": "api",
  "settings": {
    "temperature": 0.7,
    "top_p": 0.95,
    "stream": true,
    "stop": [],
    "frequency_penalty": 0,
    "presence_penalty": 0
  },
  "parameters": {
    "max_tokens": 4096,
    "temperature": 0.7,
    "top_p": 0.95,
    "stream": true,
    "stop": [],
    "frequency_penalty": 0,
    "presence_penalty": 0
  },
  "metadata": {
    "author": "OpenAI",
    "tags": [
      "General"
    ],
    "default_max_tokens": 2048
  },
  "engine": "openai"
} 

c) Restart Jan, and you should see the new model available!

jan_data_folder.mp4