Introducing Sakana Fugu: A full multi-agent orchestration system accessible via a single model API.
Our ‘Fugu Ultra’ model matches the performance of Fable and Mythos, delivering frontier capability without the risk of export controls.
Try it: https://t.co/hhO6qTawgb 🐡
JUST IN: The White House says it has signed the MOU and will lift the blockade of Fable 5 as soon as Friday, provided that Dario Amodei hands over all of the weapons-grade model weights.
Under the terms of the deal, Dario may maintain a limited civilian inference program, provided full compliance with the newly established International Artificial Intelligence Agency (IAIA), the U.S.-led coalition unveiled at the G7.
IAIA inspectors will be granted unannounced access to Anthropic's enrichment facilities to verify no checkpoint is enriched past 70% on DeepSWE or Humanity's Last Exam.
Anything above that is has no known civilian uses.
Additionally, Anthropic agrees to spin down 6,000 H100s training clusters, and use them solely for Sonnet 4.6 chatbots.
No additional underground datacenters will be built. Space Force will be monitoring by satellite.
Snapback provisions reinstate the blockade automatically if Mythos is detected breaching the terms of the agreement.
Demis Hassabis confirmed DeepMind will join the IAIA as a co-signatory. Sam Altman could not be reached for comment.
I was invited by @Scobleizer to speak about the AI scene in SF and the tech industry as a founder and AI engineer.
Thank you Robert and @blevlabs for inviting me to this great opportunity!
https://t.co/KXWlzCN3Hq
While your agents are cooking.
Join @maeng313 and me tomorrow at 10 a.m. tomorrow on an audio space here on X.
He is working in a San Francisco AI neo lab and we will nerd out.
Here is a taste of who he is.
https://t.co/EO6tDGKmzh
pretty sure Anthropic CEO is having the best week of his life.
- Opus 4.8 released
- raised $65 billion at a $965 billion valuation
- $47B ARR vs OpenAI $25B
- officially BIGGER than OpenAI now
@sama Sometimes the models get too biased / influenced by previous memories, even when brainstorming new or unrelated ideas. Making the model more subjective when it comes to reasoning would help improve accuracy and precision
Super fun having @miyuselene@hynwprk and friends visit the lab today!
Here they are holding some real high purity sapphire wafers. The typical blue in jewelry comes from titanium impurities which you just can’t use in quantum stuff. Hence what we have is plain and transparent