Would it be alright if I translate it into Korean and share it with the Korean AI community? I previously translated your DeepSeek article—here’s the link:
Love your content on AI! I'm Sia from Novita AI—we help developers access and deploy LLMs instantly, without the hassle of managing infrastructure themselves.
We're currently building our creator network through an affiliate program. Your followers are exactly the kind of developers and builders who benefit from our service, and I think this could be a valuable opportunity for you.
The "six years" framing in the subtitle is doing a lot of work. What it's pointing at is a specific moment: 2019, when OpenAI released GPT-2 in staged phases, framing it as "too dangerous to release in full." That was the moment the lab committed to a closed-by-default posture that lasted until now.
What's interesting historically is that this mirrors a much older cycle. In the early 1970s, software was treated as something you bundled with hardware and gave away — Xerox PARC shared freely, Bell Labs distributed Unix source code to universities. Then in 1976 Bill Gates wrote his Open Letter to Hobbyists and the commercial software era began. Open and closed have traded dominance several times since.
The fact that releasing an open weights model in 2025 is newsworthy suggests we've been in a closed era for long enough that the default has shifted. Worth watching whether it shifts back. I cover the long arc of these cycles at The Long Compile.
Hi,
This is Michael Shin.
I really enjoyed your article!
Would it be alright if I translate it into Korean and share it with the Korean AI community? I previously translated your DeepSeek article—here’s the link:
https://tulip-phalange-a1e.notion.site/DeepSeek-R1-189c32470be2801c94b6e5648735447d?source=copy_link
Thank you!
Best regards,
Michael Shin
Please do! That would be great! Thanks for translating the DeepSeek one!
Done!
It was fun translating!
Thanks for the article!
Here is the link for the Korean translation:
https://tulip-phalange-a1e.notion.site/GPT-OSS-255c32470be2805dbc87c2172d697de7
Thanks for sharing, Jay.
Hi,
Love your content on AI! I'm Sia from Novita AI—we help developers access and deploy LLMs instantly, without the hassle of managing infrastructure themselves.
We're currently building our creator network through an affiliate program. Your followers are exactly the kind of developers and builders who benefit from our service, and I think this could be a valuable opportunity for you.
Happy to share details if you're interested.
Best,
Sia
The "six years" framing in the subtitle is doing a lot of work. What it's pointing at is a specific moment: 2019, when OpenAI released GPT-2 in staged phases, framing it as "too dangerous to release in full." That was the moment the lab committed to a closed-by-default posture that lasted until now.
What's interesting historically is that this mirrors a much older cycle. In the early 1970s, software was treated as something you bundled with hardware and gave away — Xerox PARC shared freely, Bell Labs distributed Unix source code to universities. Then in 1976 Bill Gates wrote his Open Letter to Hobbyists and the commercial software era began. Open and closed have traded dominance several times since.
The fact that releasing an open weights model in 2025 is newsworthy suggests we've been in a closed era for long enough that the default has shifted. Worth watching whether it shifts back. I cover the long arc of these cycles at The Long Compile.
Hi Jay,
My name is Swarup, writing from India. I’m a big fan of your blogs and your book.
I would love to translate this particular blog post into Hindi and Odia (Indian Languages).
If the effort works out well, I’d be happy to translate additional posts as time allows.
Please let me know if you’re comfortable with that.
Thank you,
Swarup
Please! Go for it!
Hello, Jay.
I am a big fan of your blogs, which helped me a lot at better understanding LLM.
Recently I translated this post into Chinese and posted it at Zhihu, which is a popular Chinese knowledge-sharing platform.
Here is the link: https://zhuanlan.zhihu.com/p/1943201595404813410.
Thanks for sharing.