Combine the world's best AIs to create most beautiful answers you've ever seen.
We're so happy you've been a loyal user! Ithy has only existed for a few months, so we're very glad you've been along with us for most of our journey!
It's definitely changed a lot since our first beta release, and today's official 1.0 launch on Product Hunt is a huge milestone for us...thanks for being here to support Ithy on launch day!
Hi hunters, today I'm excited to introduce the launch of Ithy.
✨ Ithy combines the world's best AIs, blending their answers into a single response.
Ithy is more than just a chatbot. Every question is answered with a comprehensive article: complete with pictures, videos, charts, and other insane visualizations. Experience the world's first multimodal AI at ithy.com
Highlights:
Current models include Gemini 2.5 Pro, OpenAI GPT 4.1, Grok 3, DeepSeek R1, Perplexity Sonar, and many more. We don't just show their responses separately - we unify them into a single incredible answer.
Every answer is crafted into a fully functional HTML webpage (complete with pictures, videos, charts, etc.).
We've created the world's #1 richest multimodal AI answers, and the world's only interactive AI answers: play videos, interact with data, and quiz yourself with FAQs.
Up to 10x faster than Deep Research with our proprietary technology (Asynchronous Mixture-of-Reasoning) and over 200% more sources than typical AI chatbots.
Here's an exclusive Product Hunt invitation to our Discord community - be part of our new journey!
And our latest research and product releases are always shared on X, so follow us on X to be the first to hear our updates.
Just added to our free tier:
✅ Gemini 2.5 Flash Thinking
✅ GPT 4.1 Mini
✅ Grok 3 Mini
✅ ...and more
And now in our Pro tier:
⭐ DeepSeek R1
⭐ Gemini 2.5 Pro
⭐ OpenAI o4-mini
⭐ Grok 3
Combine them for some INSANE results:
@wluk - this looks dope, not going to lie I have often done this manually, specially when it comes to creative tasks like writing an email or an article. Love your approach to this, once you get the responses from the various LLMs, which model are you using to curate the final output - if it is a model / else are you doing that programmatically?
Cheers - all the best!
Thanks for the awesome words@parth_lohomi! You can think of Ithy as an engine, not a model. It's a very complicated system that combines all the different models intelligently, but you're correct that the final output is written with multiple LLMs, as well as a lot of programmatic logic.
@wluk ahh super cool - all the best!
This is wild – blending Gemini, GPT-4, Grok, and more into a unified answer? The multimodal aspect + interactive responses make this feel like the future of AI search. Huge congrats to the man!
Thank you for the encouragement @suryansh_tiwari2! I know it sounds ambitious, but our goal is to be the only AI you'll ever need to use!
@suryansh_tiwari2 We're actually one of the only AIs to offer GPT-4.1 (not just GPT 4) AND o4-mini, which means we're able to combine the best models from OpenAI with the best models from Google, Grok, DeepSeek, and others.
I tried it right away and I have to say: incredibly wonderful and incredibly... terrifying at the same time ;) In a minute Ithy created not only an article, but a complete page with photos and an embedded video! Ok, it made a few mistakes, but the prompt was also quite specialized - it concerned the regional cuisine of Upper Silesia, which even "living" Poles have problems with ;) And that's wonderful.
It's terrifying that I, as a writer, journalist... can I die now, right? ;) People are unnecessary, since (everything?) an app will do it for us?
Yes, exactly! However, we - people - are needed to catch such errors and correct them. Add to them... our own flavor ;)
@new_user_afd351ebdb If you share the article here, we can all take a look!
But if you're a writer/journalist, I respect that a lot, and that's something AI will never replace! The hardest part of your job is deciding what the most important issues are to talk about, as humans.
@wluk thank you :) And sory, yes, of course, here is the generated article: https://ithy.com/a/rw54120k3k
@new_user_afd351ebdb Awesome! This is the first Polish article I've seen on Ithy, and it's great!
Very satisfying watching the different models generate their responses live! It almost feels like the model has multiple parallel reasoning chains going on, instead of just one chain of thought like other models
Pretty cool! I use PPLX pro as my daily driver. One feedback would maybe be to condense the answer a bit. Its a bit long in the tooth. Or have the user be able to some how set length.
@sentry_co I agree too! That's why we made a new mode called Fast Research. Try switching to the "Fast" mode - the responses are a bit shorter, and a lot faster.
@sentry_co Try putting that in the "Custom Instructions" section then! You can specify output length, output format, tone, etc.
I've seen other multi-AI systems but the execution on this is really impressive. As a test to the free models I wanted to see how it would handle a well documented by often tricky topic (a specific set of doctrines from a specific sect of Christianity) and it absolutely nailed it, providing not just the direct answer, but historical information, and some really interesting mind-map, data graphs, Q&A, and tables I was not expecting. Signing up for the Pro version to take it for a deeper spin. Hats off on your efforts, and may your servers hold the line. (and your wallet as you have to be losing your savings in this kick-off...?)
@1lastshot Thank you for the support Brett, it means so much! The best way to test a new system is to ask it questions you already know the answer to, so I'm glad Ithy passed that test!
We just want to give the best answer possible, cost notwithstanding ;)
Why is it trying to open my Metamask wallet? @wluk
@distancerun_ I think there was a scam token that copied our name and website. We are not affiliated with any crypto projects.
Ithy
Thanks for the support Don! My goal is to make Ithy better (or at least different) than everyone else, so I really try to make it give the most comprehensive response every time.