how soon until AI companies figure otu that enterprise users will not be a cash cow they all think they are.
Paul S retweeted
gpt-6.1 terra on the first prompt vs gpt-6.1 terra after 3 more
same island
same pagoda
same dragon
one is a sketch, the other is a world
the model isn't lazy. your first prompt is
Paul S retweeted
OpenAI reducing limits on October 30th will be a disaster for the company.
I am SHOCKED at how bad the limits are already, I'm at 12% left with 5days left until the plan renews. This will cause a mass exodus from Codex.
If they do follow through on this plan, once my resets are done, and my one tie allotment of 62500 credits are gone, i'll be shifting my Pro plan to Claude Max.
Paul S retweeted
Day 23: a usage limit reset...
A reset is nice, but let's not forget that ChatGPT Pro limits are being cut from 20x to 10x in less than 3 days. This hits existing Pro $200 subscribers who kept the old 20x allowance, and yes, they got a one-time 62,500 credits (about $2,500) to soften the blow, but a one-time grant doesn't make up for a permanent cut.
There are only 5 days left in Tibo's 28 Days, and I really hope OpenAI uses them to fix this, whether that's rethinking the cut or shipping a model that's as good as Opus 5.5 while being way more efficient than GPT-6 Astra.
Paying users deserve more than half of what they signed up for.
Paul S retweeted
Gpt 6.1 Terra高速通道体验服务已到期
目前 TPS 已回落到 50 左右。。。不排除进一步下降的可能
另外 TTFT 速度已经到了 20S 左右。。。
各位如果觉得慢了，不要骂娘，因为还是比 Gpt 6.1 Sol要快的
Paul S retweeted
GPT 6.1 Terra is a mode that really benefits from multiple iterations tog et great results.
Here is a comparison between the first prompt and the fifth iteration
The improvement is pretty impressive.
Paul S retweeted
If Claude wrote your website's privacy policy, it may have handed 𝘆𝗼𝘂𝗿 𝘄𝗵𝗼𝗹𝗲 𝘀𝘁𝗮𝗰𝗸 to attackers.
Someone found that when Claude writes a privacy page, it lists every third-party tool you've plugged in and adds promises you never meant to keep. Press it and it says it writes from your customers' side by default.
A page that spells out which services you use hands attackers a checklist for known exploits, phishing, and credential stuffing. Send this to Claude Code or Codex to pull the traps out of your policy pages first:
"Read my privacy policy and terms pages. Quote every promise, guarantee, named person, and vendor, and list them for me to confirm. Swap tool names for categories (hosting, payments, analytics). Change nothing until I approve."
Paul S retweeted
Impressive
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
Paul S retweeted
i had a chance to try Argon before joining deepmind but after using Carbon internally I can tell you that this team is absolutely cooking and we’re only just getting started.
Carbon is writing all of my code and I can trust it to run all night, for multiple days, making intelligent decisions autonomously.
there is still lots of work to do so we’re excited to get this in your hands and hear your feedback.
we can’t build AI for everyone without the community 🖤
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model.
Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching.
We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards.
As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
Paul S retweeted
Gemini 4 Carbon is our next frontier model.
It’s state-of-the-art across cybersecurity and long-horizon capabilities.
sol 6.2
Plot twist: maybe the next OpenAI win isn't Astra.
Someone asked Tibo for a stronger Sol that can compete with Opus on speed and cost.
His reply? "Taking notes."
Two hours later, he teased "good news" for the next shipping day.
Could be completely unrelated, but I'd love a faster, more capable daily-driver Sol.
Would you rather get that or Astra 6.1?
Paul S retweeted
The Pro 200 usage cut is in a few days.
I was on 20x but am now on 10x and I have some insights.
It's MUCH worse than I thought.
Don't even think of using 6 Astra, your weekly can be killed in a literal hour if you don't realize.
Even normal coding use on the "cost efficiency 6.1 sol" can use up to 50% of the weekly with normal coding work.
I don't even use the codex sub as the main, I use Claude 20x and opus 5.5 as my main agent and only use 6.1 sol for reviews and a bit of implementation and the usage limits are HORRIBLE.
Don't forget that Tibo's 28 days shipping challenge is ending soon so say bye to your resets.
terra is an iterative multi step approach, its getting close to the same concept as opus 5.5.
용 한 마리를 만드는 데 무려 11시간 35분 45초가 걸렸습니다... 🤯
--> 어제 새로나온 GPT-6.1 Terra (MAX)에게 시켰고요!
(블랜더 사용 X, 오직 three.js으로만)
처음 만들어달라고 했을 때는 진짜 처참했습니다.
이게 용인지.... 🦇 플라스틱 박쥐인지 구분이 안 갈 정도였습니다.
그래서 이번에는 조금 다르게 해봤습니다.
아예 Goal로 걸어놓고, 용 사진 하나 던져두고...
--> 이 사진 정도의 퀄리티가 나올 때까지 계속 반복해서 수정하고 개선하도록 시켰습니다.
그랬더니.. 무려 11시간 35분 45초 동안 작업이 한 번도 끊기지 않고 계속 작업했습니다.
그렇게 수없이 수정하고 개선해서 만들어낸 최종 결과물이 바로 이 영상입니다.
🐉 하늘을 날아다니는 용의 모습부터 날갯짓, 🔥 불을 뿜는 모션까지!
확실히 goal로 사진 한 장 던져주고 계속 무한 개선 시키니까..
훨씬 더 잘 만들어주긴 합니다....
쨌든 이번에는 원샷 프롬프트로 만든게 아닌 점!
고려해 주셔서 봐주세요!