Hi folks,
One thing that has bothered me for a while is that LLM answers can sound confident and still be wrong.
To deal with that, I originally wrote a small script for myself: send the same prompt to multiple LLMs, then use another model to synthesize the responses into a final answer. I found this workflow surprisingly useful, especially when I wanted more confidence in the result instead of trusting a single model too quickly.
Recently I spent a bit of time turning that workflow into a small iOS app and made it free.
A few things about it:
It supports multiple mainstream LLM providers. You can configure them with your own API keys.
The goal is not “model comparison” for its own sake, but reducing hallucinations and making answers easier to cross-check.
The app is lightweight — currently under 1 MB.
It is designed to send your input directly to the configured LLM provider rather than through my own backend.
If anyone is curious, it should be easy to find on the App Store by searching teramoby.
I’d be happy to hear any feedback, especially on whether this kind of multi-model workflow is actually useful in practice.
Top comments (0)