
Staff Engineer | OSS @t3dotcodes | Available for software contracting
Joined April 2008
- Tweets616
- Following132
- Followers124
- Likes325
Used @t3dotcodes with Fable 5.2 to coordinate the session, find targets, improve focus and framing, and crop for the diagram in the post
First photos from my new @Unistellar eQuinox 2 smart telescope! Thrilled with how they turned out #Astronomy #Unistellar
7
First photos from my new @Unistellar eQuinox 2 smart telescope! Thrilled with how they turned out #Astronomy #Unistellar
I like the accessibility wins of using @t3dotcodes with a gamepad and dictation in a fun setting
1
7
Mike Olson retweeted
This is what it looks like when you turn on subsidization, btw
Replying to @cognition
On our price vs performance evaluation, models cluster into several groups.
▪️ Opus 5.5 and GPT-6 Astra lead in intelligence, at average costs of $12.8 and $6.67, respectively.
▪️ GPT-6.1 Sol leads in cost efficiency, scoring 39% at $1.03 per task.
▪️ GPT-6 Luna and Haiku 5.5 lead on overall cost, scoring above 20% at less than $0.50 per task.
Chinese OSS models lag far behind, with Kimi K3 and GLM 5.3 scoring at 17%. American OSS models like Beam and Inkling are even further behind with a score of 3%.
Mike Olson retweeted
Many of you don't understand how serious this is, so let me break it down.
In computer science there is a famous kind of puzzle called 3-SAT. You are given a long list of yes-or-no switches and a set of rules about how they must be arranged, and you have to find a combination that satisfies every rule. The obvious way to solve it is to try every combination, but each new switch doubles the number of possibilities, so a puzzle with a few hundred switches would take longer than the age of the universe to brute force.
For decades, the smartest people in the field found clever tricks that made this a little faster, but every trick still hit the same wall: the time required kept exploding as the puzzle grew.
About 25 years ago, two computer scientists, Russell Impagliazzo and Ramamohan Paturi, working with Francis Zane, turned that wall into a formal idea.
They proposed the Exponential Time Hypothesis, which says this explosion is not a failure of human cleverness but a law of nature. No algorithm, however brilliant, will ever solve 3-SAT without the time blowing up exponentially. Nobody could prove it, but almost everyone believed it, and it became one of the foundations of modern computer science.
Researchers used it as a measuring stick. If you could show that solving some other problem quickly would also crack 3-SAT quickly, you had shown that problem was hopelessly hard too. Hundreds of papers, across scheduling, graphs, logic and much more, rest on this one assumption.
Now an internal AI model at OpenAI claims to have broken it. It says it found an algorithm that solves these puzzles meaningfully faster than the hypothesis allows, and it has checked the proof in a computer verification system called Lean.
And here is the remarkable part: the very people who proposed the hypothesis, Impagliazzo and Paturi, along with MIT's Ryan Williams, are now leading the public review of whether a machine has disproved their own idea.
So what happens next? If the disproof holds, a large part of theoretical computer science has to be revisited.
Every result that said "this problem cannot be solved faster, assuming ETH" loses its foundation, and researchers will have to work out which of those walls were real and which were simply where we stopped looking. It does not mean everything becomes easy. The new algorithm is still slow in absolute terms and this is not the famous P versus NP problem being solved.
But it means our map of what is possible was wrong in a place we were very confident about, and it was an AI that redrew it. If the disproof fails, the hypothesis survives and the field breathes out, but we will still have watched an AI produce serious, checkable work on one of the deepest questions in the discipline, closely enough that the founders of the idea had to take it seriously. Either way, something has shifted.
After seeing what @theo is doing for rust and tsc-rs, I’m curious about whether Opus can hill-climb efficiently on aube to get it compatible enough with pnpm 12 to build t3 code, and have aubeshim use it to finally replace all of {npm,pnpm,bun,yarn} on my system
1
1
23
I still can’t get used to seeing t3code next to VS Code and webpack in a Cloudflare post; some neat collab happening here
Introducing tsrs: a Rust port of the TypeScript compiler 🦀
We are not forking TypeScript!
◆ In close collaborating with the TypeScript team
◆ Up to 8x faster than TS 7.0 & ~70% less memory
◆ Runs in WASM, so it works in your browser & on Workers
◆ Soon part of the Oxc toolchain
npx tsrs
blog.cloudflare.com/typescri…
28
Orchestration in T3 Code v0.1.0 lets Claude, Codex, OpenCode, and Pi subagents work together in one thread. Here's what the latest @t3dotcodes release does and which parts I worked on:
mwolson.org/blog/2026-10-15-…
1
2
93
Mike Olson retweeted
Orchestrator V2 is now out on the stable channel in @t3dotcodes! We also thought it was time we drop the Alpha branding, so we call this release 0.1.0 Beta.
Mobile store builds are also updated. Versions 2.0.0 are available to everyone on both App Store and the Play Store!
One of the more rewarding parts of working on aubeshim is being able to get aube working in projects that aren't using it at the moment, and identifying 28 different problems with repro cases that helped them get fixed.
aube has a very nice workflow where I can file discussions for a maintainer's agent to look at, and has been one of the easiest projects to contribute to.
github.com/mwolson/aubeshim
5
Huge improvements for subagents while still starting quickly! Next release will be the one to try out if folks haven’t yet.
13
Mike Olson retweeted
The thoroughness with which agents can test and QA software is unrivaled. I pointed out something wasn't rendering right in a mobile (in-app) browser. The thing goes to the depths of Mordor to reproduce, simulate, fix, deploy, verify. It created an ephemeral vercel deployment to throw at an iPhone simulator! The absolute grind on these things.
Humans could simply never match this level of intensity. We'd run of energy, we'd extrapolate, we'd hope it'd work out. The software of the future will be of a quality and performance we've never experienced before. It's amazing to watch this shift.
Greg Brockman says OpenAI pointed Astra at its own systems until it ran out of vulnerabilities to find:
"We took 25% of our production engineers and said, 'Sorry, all your projects are on hold. You are now defending. You are now up-leveling our security architecture. You're going to use the models to find all the holes.' And we found a number of serious issues, and we fixed them."
"We found some new problems, but eventually it saturated. We basically have found, to our knowledge, all of the P0s, all of the critical problems that Astra is smart enough to find. And of course, there will be a new model, there will be a new round."
"You want to be in this tight loop of new cyber capability drops, you deploy it against your systems, you find the new holes, and ideally, you've managed to automate this, what we call defense factory. That's what we're building internally."
"There are ideas, for example, formally verifying all of software, that are possible with AI."
@gdb @bhorowitz
Greg Brockman: "We're now in the AGI era."
Ten years ago, OpenAI worked out the compute curves and landed on fifteen years to AGI, or ten if the world was willing to build the machines and spend the hundreds of billions to do it.
In 2026, GPT-6 Astra manages 24 hours of coherent operation, 10,000 agents worked together to solve Navier-Stokes, and a model ingeniously chained together exploits to break containment at Hugging Face.
@gdb joins @bhorowitz and @eriktorenberg on what the AGI era asks of us: why safety and alignment now set the pace, what happens to work, and why AI sentiment is lowest in the country building it.
00:00 Intro
00:52 15 years, or 10 if you spend enough
02:24 Pacing the frontier
03:55 Safety ideas from before the models
08:42 Lessons from Hugging Face
10:20 The defender's window
12:46 10,000 agents on Navier-Stokes
14:50 Formally verifying all software
17:10 Cancelling his holiday for GPT-3
18:42 Codex found 13 holes in 15 minutes
20:41 Why Astra earned the GPT-6 title
24:25 Employment keeps going up
29:39 America has the lowest AI sentiment
31:10 The benefits don't make the news
33:26 Banning data centers exports them
35:50 $1 billion for frontline defenders
38:15 Astra cleared the bar for AGI
40:18 1.5 billion people churned ChatGPT
43:25 Killing Sora
47:45 The AGI era
YouTube: youtube.com/watch?v=IJn8cagM…
My pi-xai-ws Grok adapter (uses WebSockets) for @pidotdev now has a v1.0.0 release with great cache coherency.
Over my own use of 2,787 Grok Responses calls, 672 million prompt tokens:
- 96.1% of the tokens were cached
- 99.5% cache rate on the median Grok Responses call
mwolson.org/blog/2026-09-09-…
1
20
pi-xai-ws now has significant performance, caching, and error-handling improvements for Grok models in Pi.
Grok models can now be reached in Pi using the official WebSocket interface with my extension pi-xai-ws pi.dev/packages/@mwolson-org…
21
Mike Olson retweeted
I just set up Grok @Bot to remotely control my autonomous robot lawn mower that is 50 miles away. It literally took 2 minutes to set up.
I told it to start mowing and it started 3 seconds later. Then I had it dock itself. This is so wild lol.
I'm gonna set up my vacuum next, and have a Grok Bot dedicated to controlling and managing all of my autonomous robots.
Grok models can now be reached in Pi using the official WebSocket interface with my extension pi-xai-ws pi.dev/packages/@mwolson-org…
























