I write about tech. Sr. Editor, AI at The New Stack. Signal 860-208-3416 [email protected]
Portland, OR
Joined March 2007
- Tweets19.5K
- Following4.1K
- Followers25K
- Likes7.1K
How many pints did he have before he said that?
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
Frederic Lardinois retweeted
In lieu of flowers, memorial contributions may be made to:
Johns Hopkins Alzheimer’s Disease Research Center Fund
alzresearch.org/giving/
Alzheimer’s Foundation of America Caregiver Resources
alzfdn.org/donate/
Photo Credit: Tommy Garcia
Frederic Lardinois retweeted
Obama tonight on his Cheesecake Factory line: "They mobilized a whole of government response to that joke. Not to inflation, but to that joke. The White House even called the Secretary of the Treasury. This is the man who is in charge of the economy. The person who's supposed to be figuring out how to make your life more affordable. The guy who's supposed to be figuring out how to lower inflation. And reduce interest rates. They told him he should hold a press conference in front of the Cheesecake Factory."
Frederic Lardinois retweeted
We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent.
It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.
Frederic Lardinois retweeted
I still can’t believe that United wants you to spend $32,000 on flights for one year of Premier 1K status.
That’s more than many Americans take home all year. Our economy is wild.
Frederic Lardinois retweeted
Meet Nano Banana 2.1, our latest image generation and editing model 🍌
This upgraded version outperforms our previous models across the board, with notable leaps in visual design, mask-based editing, and subject consistency to help you create more natural-looking images.
Frederic Lardinois retweeted
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
🤖 Made with AI
Frederic Lardinois retweeted
Claude now works inside Google Docs, Sheets, and Slides, and those files also open inside Claude.
In Google Workspace, Claude sits in a sidebar next to your file, reads what you have open, and edits it in place. You can approve each edit before it lands.
"Super intelligence" is now Trump-coded, which is probably exactly what he wants. I'm not going to use that, of course. Not my fault when tech CEOs bend the knee to Trump.
Frederic Lardinois retweeted
Claude Code will now try to find a graceful stopping point when you hit your 5-hour limit mid-task, instead of cutting off mid-edit. It gets a small, fixed allowance pulled from your weekly limit to wrap up what it can.
Frederic Lardinois retweeted
I've been launching models at OpenAI for a while now, and GPT-6 Astra was the first launch where the model actually did the majority of the manual work that I usually do, while I focused on the strategy, judgement & human-facing work.
For this launch, GPT-6 Astra in ChatGPT:
- Built and maintained our media list in Google Sheets
- Kept our comms plan updated as the team executed
- Drafted and sent pitches and managed embargoes
- Tracked reporter replies and questions, and sent clarifications
- Managed press briefing RSVPs in real time and sent me hourly updates
- Created a branded PDF of the blog and checked it against the final copy and evals
- Collected assets and built the press kit
- Tracked coverage in real time, flagged inaccuracies in Slack to the right person, and drafted replies requesting corrections
- Built a ChatGPT Site to track and analyze coverage, social reaction, and key metrics
- Wrote our coverage report using our team's template and skills
Meanwhile, I:
- Focused on the narrative and strategy
- Kept my team aligned
- Called reporters on the phone
- Prepped our spokespeople
- Hosted the press call
- Staffed live broadcast interviews
- Monitored the situation
- Got a full night's sleep before launch (!)
- Kept up with all the usual last-minute changes without the usual stress
If this is the AGI era, I am here for it! Excited for the world to get to experience GPT-6 Astra. It's a very good model and I think that it will transform both the way we work and how much we can accomplish.
It’s almost like this president doesn’t know what he’s talking about. axios.com/2026/08/23/trump-d…
Frederic Lardinois retweeted
Last month you told us Remote Control was the thing you'd most like us to fix, so we've been working hard on reliability.
Here's what's better today: 🧵
I spent $0.50 on OpenRouter today, so I think that purchase price is completely justified. nytimes.com/2026/08/19/busin…
