Self-reported, but Terra seems to outperform both latest Astra and Sol in some benchmarks at half the cost of Sol. Well done @OpenAI
Replying to @OpenAI
GPT-6.1 Terra substantially outperforms GPT-5.6 Terra on Agents’ Last Exam, a test of how well agents use software at work.
It even matches GPT-6 Astra at one-tenth the price per token and finishes tasks over 3× faster on average than both Astra and GPT-6.1 Sol.
Art retweeted
I got to spend 30, 40 years writing code, mostly video games, and I'm just really grateful for that experience. What a dream.
I loved every little piece of it.
The bugs you'd fight with for days. Going to bed thinking about code and waking up with the answer (sometimes).
That feeling when something you crafted and agonized over finally just works. And doing it next to artists, testers, audio/musicians, animators, producers, designers and programmers, shaping these experiences together.
Now, in less than a couple of years, it's mostly AI writing the code, and it's bittersweet.
Looking back, that was a golden era, kind of a golden opportunity in the whole history of humankind, to program these things by hand. I don't think it comes back.
But I also think the next 30 or 40 years are going to be incredible for how these things get created and shaped.
So, grateful for the opportunity. A little sad. Very excited.
Art retweeted
LOL my agents got themselves banned from OpenAI autonomously 🙃
haven't touched this account myself in days so it had to have been from the cron jobs
what did the little rascals get up to??
Art retweeted
GPTでBAN祭りになっている模様。自分も気を付けたい。
驚いたのがcomputer useで画面キャプチャなどのローカル情報を読み取られBANされる可能性。通常は禁止行為を繰り返したり迂回を試したりするとBANを食らうようだが、一線を越えると一撃BANとなるらしい(特に1)と2))。
1) 未成年エロ(例:JK)
2) 犯罪関与
3) jailbreak (LLMをイジメ抜くのも含む)
4) センシティブな技術アクセス(核・生物・クラッキング)
5) 契約上の問題(契約者とクレカが一致しないなど)
This bug is biting again. @OpenAIDevs @thsottiaux just got my account banned without warning or rejections in codex.
Normal TS+Swift dev, no Claw/Hermes.
Appeal auto-rejected after 60min.
An issue caused some user accounts to be incorrectly suspended.
We’re restoring access and working through related subscription and credit issues.
status.openai.com/incidents/…
Wait, 5.6-Luna outperformed 6-Sol in this benchmark? 🤔
If Astra is as good as the influencers are painting it at computer use + visual reasoning, it might be a big white-collar job booster.
Most AI deployments are slow b/c of having to work with legacy systems. This could make it irrelevant...
Can't wait to test Astra on a difficult infra task, that 5.6-sol(max) has been struggling with for days...
GPT-6 Astra is here.
We hope it will begin to enable a new generation of entrepreneurship, scientific discovery, and building.
We believe it is the best model in the world for computer use, professional work, science, coding, cybersecurity, and more.
It took us some extra time to ensure that we could meet the safety and alignment standards required for this capability level, but we think you’ll find it worth the wait.
It scores 98% on FrontierMath Tier 4, 99.9% on ARC-AGI 3, and 100% on ExploitBench.



















