Difficulty configuring multiple instances of the same model with distinct parameters

[mmtpo]

I am currently self-deploying an application that requires setting up multiple instances of the same model, each configured with different parameters. For example:

MODELS=`[{
      "name": "gpt-4-0125-preview",
      "displayName": "GPT 4",
      "endpoints" : [{
        "type": "openai"
      }]
},
{
      "name": "gpt-4-0125-preview",
      "displayName": "GPT 4 temp 0",
      "parameters": {
      "temperature": 0.0
  },
      "endpoints" : [{
        "type": "openai"
      }]
}
]`

This results in a state which looks like that both models are active simultaneously.

However, in practice, I cannot activate the second model ("GPT 4 temp 0"); only "GPT 4" is utilized during chat operations. It appears as if the system defaults to the first model instance and ignores subsequent ones with the same model name.

I tried to distinguish between the models by modifying the name field and introducing an id field, using the appropriate model identifier. However, this approach resulted in a loss of model reference, indicating that these fields cannot be arbitrarily configured on the client side.

Is there a recommended approach to deploying two instances of the same model with varying parameters? Any guidance or suggestions on how to achieve this would be greatly appreciated.

[wdhorton]

I've also run into this