Model Hub

This is a place that you can compare different models, to see how they perform, given the same input. There are 1 or 2 copy websites I have seen in recent month, but we have really been one of the few to ever use this or have it working. Under appreciated for people who actually want to use AI to get results.

Something we did not clearly state, is that the Model hub is pretty much just for testing the BASE models. This means no internet connection, all the models have available to them is the chat, their weightings (training), and any other inherent characteristics e.g. some models have add-ons for describing images.

So do not expect them to perform the same as they do on the parent companies website, as they have been stripped down of all the smoke and mirrors, to show the knowledge base of the model itself. Model hub -> RAW base model. Just knowledge base. Agent Hub -> Can interact with other tools and structures e.g. memory, reasoning, internet access, max output can be different. This can make it hard to know if a response is better or worse between models, as the tools, the prompting, and all the little extras, play in different ways. Difference in wording even, can make some models perform better than others.

This is all to say, the model hub is supposed to give you a "raw" base model experience.