Quick takes on new tech & AI. Long-form on YouTube ↓
Joined July 2022
- Tweets131
- Following24
- Followers4
- Likes339
Pinned Tweet
HyperContext is all about WHY?
To answer “why” you need more context.
That’s HyperContext.
1
1
28
Google just shipped Gemini 4 Carbon: $2/$10 per 1M tokens, #1 on three agent benchmarks, and a "/swarm" mode where multiple agents split up a task.
The catch: the price doubles on Nov 30. Models keep getting better, and the intro-price window is now part of the playbook. If you're building on it, test now.
1
4
Australia just accepted all 18 recommendations from its rapid AI review.
Frontier AI labs there will have to report serious incidents fast, answer to a new safety regulator with real penalties, and meet six mandatory security obligations.
Whether you like it or not, "ship first, explain later" is getting a legal expiry date.
1
4
Claude's NEW Fable 5.2 seems to perform great on benchmarks. It almost leads every category and is a significant upgrade to Fable 5.1.
1
10
Claude Fable 5.2 is here.
Anthropic’s new frontier model targets complex software engineering, long-running projects, and agent coordination at scale while generating output up to 30% faster and costing 20% less than Fable 5.1.
Now $8/M input tokens and $40/M output tokens.
Fable 5.2 is generally available today.
There are some offers for eligible plans.
1
14
Something feels off about the timing.
Democrats just swept a bunch of states in the midterms. And right around the same time, OpenAI and Anthropic have both gone quiet. No major model drops in weeks, after a year of nonstop one-upmanship.
Maybe the quiet is nothing. Labs slow down sometimes, elections are elections…
But it raises a real question: how much of the AI race's pace is actually about the tech being ready versus the political and regulatory weather the labs are flying through?
1
4
Jensen Huang is reportedly being investigated by Nvidia's board over a relationship with a woman U.S. intelligence believes may have been tasked by Chinese intelligence to get close to him.
Forget the scandal part for a second. Think about what it actually means:
The entire AI industry's chip supply runs through the judgment, security, and personal life of basically one man.
We spend so much time pricing in export controls, chip bans, Taiwan risk, fab capacity... and almost zero time pricing in "what if the guy at the top makes one bad personal decision."
That's not a hypothetical anymore.
Key person risk might be the most underrated risk in AI right now.
3
HyperContext retweeted
Apple has another major AI gap to close: Computer Use. OpenAI and Anthropic already have agents that can operate software, click buttons and complete tasks on your behalf. Apple controls the Mac, iPhone and iPad, but still lacks a comparable capability. Why that needs to change-
NEW in Power On: Apple eyes autonomous, agentic Siri as early as next year to rival ChatGPT Dots, Meta Muse, Grok. The goal is to move beyond responding to commands to an assistant that can proactively manage tasks and get things done on your behalf. bloomberg.com/news/newslette…
Let’s see. I already use it on my Mac and I love it.
It would be so great if @raycast would have a low cost tier that would just have super basic tools like sync and dictation. All I need…
1
26
Hermes Agent looks very promising so far.
I set it up on my raspberry pi and linked it to multiple telegram bots.
I hope it proves as a great tool. But I’m confident it will.
2
I was thinking about this quesiton. Check out: youtu.be/E1n_Mu7NhaE
Should YOU HATE AI?
More and more people are starting to hate AI. I use AI every day an...
youtube.com 1
5
Check out answers and questions: youtube.com/@hypercontextyt
1
16