oMLX model switching can get stuck with Hermes
oMLX does not smoothly unload the old model and load a new one when Hermes asks for a different model. Pi handles the same situation a little better: the first request for a different model s an error, but a second try unloads the previous model and loads the new one.
With Hermes, the appears to get stuck. handles this kind of model ing more gracefully.
Key points
- oMLX may fail when Hermes asks to from one model to another.
- Pi can recover on a second try after the first model- request errors.
- The same retry behavior is not reported as working with Hermes.
- handles unloading one model and loading another more smoothly.
- A single-model Hermes setup may be safer than automatic ing for now.