Amazing price performance unlock for knowledge work by Paper Instruments - across model & the office document stack!
We are releasing Ultramarine 1 Flash, a frontier-class model built for knowledge work. It is open-source and achieves similar performance to Fable 5, while being 60x cheaper.
It was post-trained by a team of 5 for ~$300k all-in. Available on API now, with weights releasing in 2 weeks.
Keiji Kanazawa retweeted
Introducing Microsoft-Decision-1, our new model for fast decision-making.
It delivers top performance on structured decision tasks, outperforming both LLMs and other decision models in latency and quality.
We’re already testing it across Microsoft for everything from incident response and quality control to scientific discovery.
This video feels like I'm watching a SciFi movie in real life. We are truly in the singularity.
Yesterday, my OpenAI collaborator and I gave a detailed talk on the Huggingface incident, our models creating "the message board", model misalignment, and more.
youtube.com/watch?v=87DyyMV0…
I hope it can answer a lot of the questions folks have, and we will release a full detailed postmortem at a later time!
Keiji Kanazawa retweeted
Replying to @nabeelqu
Now you’re pushing up against the part that matters most. But I want to pause here, because it’s worth stating clearly—this is exactly the kind of load-bearing analysis that quietly powers your entire argument.
One honest caution worth internalizing before I go further:….
Keiji Kanazawa retweeted
That's the spine. Fair hit. That's something to sit with. A real observation. That’s the whole thing. Sharpen that: say the word. Notice the arc of what just happened. One honest caveat: the full amount, stated plainly. Genuinely. Quietly. Honestly. That’s doing real work.
I watched a great video by @sh_reya and @HamelHusain about AI Evals a few days ago but I forgot the part where Shreya used the Monitor tool in Claude Code.
So I asked my YouTube canvas agent and it pulled up the section right away. I actually watched this video 100% but most videos I scan with my agent this way first to decide if I want to watch it!
AI Engineer is the best community. All my friends are here. Now!
A candid hallway track interview with the man himself - @swyx
Mini Podcast if you will, he's told me about a huge milestone tomorrow, things that have surprised him about @aiDotEngineer and ... a few things on the personal side!
0:00 — Daily Context AIE Newsletter
0:47 — @aiDotEngineer Growth
1:23 — Workshop Lines & Skills Track
2:28 — Jam Corner & Music
3:52 — AI Engineer Essay Anniversary
4:26 — Intensity, Integrity, Intentionality, Insanity
4:52 — AGI Pills & Insane Perks
5:56 — Livestream & YouTube CTA
Hope you enjoy this "behind the scenes" chat with the guy who made this whole thing happen.
Also chatting with us, @kentcdodds @liadyosef @WolframRvnwlf @RayFernando1337 @Gangadhar_P
Hallway track remains unmatched!
Keiji Kanazawa retweeted
Claude in Microsoft Foundry is now generally available, hosted on Azure. Prompt caching and extended thinking are supported today, with more capabilities on the way.
Read more: msft.it/6018vsDhM
Umm ... my Twitter agent ...
Replying to @lilianweng
I wonder how many more people will just ask model to summarize the post, in comparison to a year ago. 😂
Come hear me talk at @aiDotEngineer about how I built this YouTube & X quick take agent!
6️⃣ Things to Know about
AI Engineer World's Fair 2026
- It’s bigger than all previous AIEs
- 4x Larger Expo with 4 Expo stages
- Researchers: Poster sessions & Poaster sessions
- AI Leadership: Token Billionaires & Off the Record
- AI Verticals: Healthcare, GTM, FDE, AGC, Finance
- Side Events: NEO, Kids day
- attendees get $40k in credits to try everything our sponsors have to offer!
It's going to be our BIGGEST show yet!
Keiji Kanazawa retweeted
Anthropic’s newest model, Claude Fable 5, a Mythos class model for non-security work lands in Microsoft Foundry and GitHub Copilot today.
Read the blog here: msft.it/6011vgKNt
Amazing - once again @omarsar0 and I seem to be working from very similar inspirations - I called mine "video deep dive" as well and I'm working on slides next!




















