if you've never seen a theo video, pls watch this one
by far the most impactful video i've seen myself, one of his bests and most important videos (imo).
very disappointed all of the model providers (opencode, openrouter, whatever)
released 'union alpha' without telling us it was a crappy model router
kudos to the cloudflare guys for telling people what they're getting
AND FUCK union alpha
I left Theo’s team two weeks ago. Now feels like a good moment to explain why.
Content creators are racing to build takes that are much hotter than any human can handle, and we may not survive this. I wanted to work from the outside to ensure the public is informed about these risks, and help the world navigate this transition responsibly.
Right now, content creators are underinvesting in safety. A creator could undergo a take explosion, or lose control of their thumbnails, without the public ever knowing. We only found out about the latest incident because the clips broke out onto the public timeline.
I don’t think that’s acceptable for a technology that might cause employment-level risks. The public should demand far more transparency. We can’t steer this content safely without more people being able to see where it’s going.
Some of this is basic: creators should disclose their progress towards recursive self-reacting, report thumbnail incidents and near-misses, meet minimum take standards, and get independent guarantees that they have actually read the article.
I’ll be joining T3 Tools to do independent evaluations of these risks. I want to show the world that these guardrails are possible, and that by doing them we can move these creators’ incentives away from racing and towards responsible uploading.
I wrote up more thoughts here on my decision and what I hope changes:
I resigned from Theo today. I spent the last three years doing pretraining research at both T3 Code and T3 Content. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. No thoughts below.
just asked this in codex re: a @t3dotchat bug where skills weren't properly showing
swapped back to t3 and there was a patch waiting for me to download
thank you @maria_rcks; please continue to read my codex sessions
Google has:
1. The worst harness of any lab
2. The worst code apps of any lab
3. The only lab with zero 3p integration option
4. The most painful bans (since it's your full Google acct)
5. The most trigger-happy ban policies
It is a legitimate risk to use Antigravity stuff until all of this is fixed.
Google/Deepmind friends, you're in last place. It's time to act like it. If you want us using your models, you need to destroy any policy that risks our accounts.
Gemini 3.8 Flash achieves 73.7% on DeepSWE.
It pushes the frontier with an impressive 8.2% increase over Gemini 3.7 Flash, at the same cost but with more steps and output tokens.