Instead of switching between providers directly, I’ve been experimenting with accessing different AI models through a single platform. For this I’ve been using Chatbot App which aggregates multiple model APIs into one interface.
From what I can tell it routes requests through each provider’s official API. Has anyone here experimented with similar multi-model setups? Curious about your experience especially around response consistency and latency across different models.
Sure. I’m sure many have.
I wrote the first AI chatbot for this very platform and from day one users of it have experimented with different models.
For me the biggest benefit of using the bigger more expensive models is that their choice of tools is usually more consistent and logical.
If you want an “all in one” chatbot then it being able to choose optimal tools to solve a user query is critical.
I also wrote and use a Signal bot on a regular basis and it also uses a large frontier model because the results are simply better and more useful and that because the number of queries isn’t that large per month it’s still affordable to run.