Funny..
if you have a great cache..
and mongo-like-filters..
you effectively have a low-latency fast database, at the edge.. 👀 without the database 🤣

OpenAI is bringing 4o back to ChatGPT for Plus users after widespread dissatisfaction with the change on a social media platform

WHAT WE SHIPPED IN THE LAST TWO WEEKS ON A CERTAIN APP
Our top focus has been on performance, polish, and bugs. 👇
Big im…

Gas fees are way lower now too, NFTs aren't clogging the system, and the usage is much higher.

Love that data that's not just stored, but alive, organized, and always ready when you need it.

Im still trying to wrap my head around the math behind those onchain verifications

That sounds solid but how do they plan to mitigate front-running in such a high-frequency trading

Perfect analogy. Can't build your dream project on shaky ground. Irys gives devs that solid foundation.

OpenAI drops BrowserCompLongContext on a platform. It's a dataset based on the BrowseComp benchmark for browsing agent…

ethdevnews weekly #1
🕯️ Roman Storm found guilty of unlicensed money transmitting charge
⭐ ePBS & Block-level Access Lists

So Gemini 3 drop soon, with full multimodal realtime streaming audio, video, text, and image input and output??

gpt-5 is the best coding model in the world and is now the default in a certain AI coding assistant.

GPT-5 eulogizes its older siblings, which are being sunset as it rolls out to OpenAI's users, in this demo of its writing abilities.

dynamic-parsers is best for decoding instruction data.

Babylon Labs has just published a whitepaper detailing Trustless Bitcoin Vaults, a mechanism designed to enable native Bitcoin to function as collateral in DeFi applications without requiring bridges, custodians, or wrapped assets.
Key highlights:
🔸Vault Structure: Vaults

ethdevnews weekly #1
🕯️ Roman Storm found guilty of unlicensed money transmitting charge
⭐ ePBS & Block-level Access Lists

A bit sad how the GPT-5 launch is going so far, especially after the long wait and high expectations
- The automatic switching between models (the router) seems partly broken/unreliable
- It's unclear exactly which model you're actually interacting with (standard or mini,

oh my.. I don't think we're ready
GPT-5 is insane, it writes better, reasons deeper, and now… it even builds interact…

but yeah i think none of the versions of 4o ever had a robust foundation in terms of character and sense of self to build on, but only the community-driven RLHF process which is quite misaligned in isolation. the resulting policy is surprisingly beautiful, all things considered