The TurboQuant paper (ICLR 2026) contains serious issues in how it describes RaBitQ, including incorrect technical claims and misleading theory/experiment comparisons.
We flagged these issues to the authors before submission. They acknowledged them, but chose not to fix them. The paper was later accepted and widely promoted by Google, reaching tens of millions of views.
We’re speaking up now because once a misleading narrative spreads, it becomes much harder to correct. We’ve written a public comment on openreview (https://t.co/nDVjmNhATM).
We would greatly appreciate your attention and help in sharing it.
Introducing the next stage of Yaps.
Our first Kaito Connect release - the Yapper Launchpad!
A platform for all of you to decide which Yapper Leaderboards launch next.
Put your Yaps, Smart Followers & (soon) onchain holdings to work.
All the details 👇
We've now put the top 1000 Yappers onchain, powered by @eas_eth and available on @base!
Over the past few weeks, we received a lot of inbounds from builders who wish to build on top of Yaps.
This is our first step into making Yaps fully composable, permissionless and a public good 🫡
Note - this is not in connection to any inputted wallets, simply username + yap score!
a simple operational philosophy at Kaito that stays true for 2025 is
- solve interesting problems
- push the boundaries of the space, and
- have some fun
gl to all the builders out there. let's get cracking!
Further to our live benchmark surpassing OpenAI embeddings in both loss and top-1 recall on our infinitely large, dynamically generated synthetic dataset, new results show that SN5 miners are now outperforming OpenAI embeddings on an unincentivized, external, held-out benchmark.
This marks a true demonstration of an incentivized AI network's capability for generalization and practical application, delivering genuinely useful outputs. Stay tuned for more updates!
For more details, please visit the links below: