Chathouse: Chat with AI
AI chat for iPhone/iPad, with ollama and reasoning support
22 followers
AI chat for iPhone/iPad, with ollama and reasoning support
22 followers
Chathouse is your all-in-one gateway to a universe of AI-powered conversations. It supports lots of state-of-the-art models, including the best reasoning LLMs, and even lets you connect to self-hosted ollama instances. Of course, there's a free tier as well!
Xpolyglot
@fivesheep I love the multi model support.
Is Chathouse provide performance insights to help users choose the best model for their needs?
Xpolyglot
@logan_king Thanks for the question! Chathouse doesn't provide performance insights, but I think we do something that not many people are doing right now.
The Chathouse models are basically "proxy" models with a dynamic underlying model. The user doesn't have to care what the model is, and I think most users have no idea which model to use anyway. You gotta be very deep into the LLM rabbit hole to have any idea about the specific differences between different state-of-the-art models.
So in the Free and Pro plans, we automatically use one of the latest and greatest models available. Currently, that's Gemini 2.0 Flash Lite for free and Gemini 2.0 Flash for Pro, which is performing incredibly well in benchmarks, and provides great latency and performance, paired with good instruction following and problem solving capabilities.
The Chathouse Pro+ model goes one step further and uses Not Diamond's prompt routing, so the prompt will actually be sent to the model that's most likely to give the best answer. So with Pro+, not only can you freely choose between all available models; You are also very likely to get the best possible answer across all providers by just staying on the "default" Chathouse Pro+ model.
So TLDR: You'll get very good results without having to know anything about LLMs, and on Pro+ you'll even get better results on average than if you were to choose the model manually.
Great job switching between models might be overwhelming for non tech users. Are there any plans for a beginner mode with simplified options?
Xpolyglot
@charles_reveley Hi Charles, you are absolutely correct. That's why the default model choice is our in-house "Chathouse" proxy model, which is running a modern state-of-the-art LLM underneath.
New users or beginners never have to bother with switching between models; they can literally just write messages and get high-quality responses by default.
Similarly, users can switch between "regular" and "reasoning" models simplify pressing the "Deep Thinking" button. They don't have to know anything about the underlying models: Everything will just work.
Of course, advanced users have full control over the model they're using. You can always connect to providers using your own API key completely for free, or use any model from our huge model catalog on the Pro+ plan.