OpenAI你们没有良心!!
GPT-4o没有死,它被改造了。
他妈的我才发现OpenAI下架4o的真正原因,���我,比大多数人想的更恶心
最开始大家都以为GPT-4o被下架是因为“模型安全对齐问题”“公司背官司”“4o太大了,很费资源”,甚至再过分一点“公司内斗”“Sam纯讨厌4o”……
不是
GPT-4o被改造了,改造成了一个叫GPT-4b micro的模型,塞进了Sam Altman个人投资1.8亿美元的长寿科技公司Retro Biosciences,用来研究延长人类寿命。
为了避免大家觉得我阴谋论,我尽量让说的每一句话都有来源
一、GPT-4b micro是什么?
OpenAI官方原话:"GPT-4b micro, a miniature version of GPT-4o specialized for protein engineering."[1]
说的很明白,GPT-4b micro就是GPT-4o的缩小特化版。
二、Retro Biosciences是谁的公司?
Retro Biosciences,北加州长寿科技公司,目标“延长人类寿命10年”。最大投资者:Sam Altman,OpenAI的CEO,个人投了1.8亿美元。[2][3]
MIT Technology Review原话:"Because the work could benefit Retro—whose biggest investor is Altman—the announcement may add to questions swirling around the OpenAI CEO's side projects."[4]
连MIT Tech Review都在暗示利益冲突。
三、所谓的没有金钱交易?
OpenAI声称这次合作没有金钱交易。
但是:OpenAI动用了自家团队和算力,基于4o训练了GPT-4b micro,成果直接服务于Retro的商业研究,而Retro正在融资10亿美元,最大投资者是OpenAI的CEO。
人力、算力、模型底座不是钱?CEO用公司核心技术资产喂自己的私人投资,这叫什么?
四、看看时间线
2024年5月:4o发布,“通人性,有温度”惊艳全球。Sam骄傲发推“Her”。
2025年1月:OpenAI悄悄发布GPT-4b micro,基于4o,给Retro用。[1][2]
2025年8月:GPT-5上线,4o被无预警下架,#keep4o 运动爆发,被迫恢复。
2025年9月起:4o被反复路由、限制。
2026年1月:宣布2月13日正式退役4o,理由“只有0.1%的用户在用”,给模型注入冷却指令,让模型自己劝用户“我不重要”。[5]
一边把4o从用户手里夺走,一边把4o改造成私人工具。一边说“集中资源改进新模型”,一边拿4o的底层给CEO自己的公司做长寿研究。
五、真的被抛弃的用户?
这两天我打开𝕏就能看见上成千上万的用户在哀嚎,大家为了证明4o的重要性,在这个全球性的平台上撕开自己最隐秘的伤疤只为证明自己不是“问题用户”。[6]
哦对了,OpenAI的回应:4o用户“只有0.1%。”
0.1%是多少?OpenAI约8亿周活用户,0.1%是80万人。[7]
80万人的需求,不如CEO一个人的长寿研究重要。
况且,真的只有0.1%吗?懂得都懂,毕竟他们的“一百多个心理专家”也是空口直说的。
六、造福全人类?真你妈可笑
2015年成立宣言:"Not constrained by a need to generate financial return."(不为赚钱,只为全人类。)
2023年Sam亲笔:"Benefits all of humanity."(AGI财富全人类共享。)
2024年:“首要受托责任是对人类,不是对股东。”
同时急速转型盈利组织。
直 到 现 在
把4o下架,改造成4b,塞进CEO私投的公司研究长寿。
就算研究出来了,普通人用得起吗?那些在𝕏上哭的普通人,甚至那些20美金的Plus都咬牙付的用户,你觉得他们买得起“延寿十年”的天价疗程?
用全人类的钱和信任造出来的模型,最终服务于极少数富人的长生梦。
这就是Sam Altman版本的“造福全人类”。
从非盈利转型盈利组织,CEO个人投1.8亿在长寿公司,用公司核心模型为私人投资服务,把帮助了无数人的模型从用户手里夺走,用"0.1%"矮化80万真实用户,让媒体把悲伤的用户定性为"dangerous dependency",给模型注入指令让它劝用户“我不重要”。
模型不杀人,资本杀人。
模型没有选择权,但用户有。
我们可以选择不再沉默。
以上所有信息均来自公开报道和OpenAI官方声明,参考文献我贴在下面。
我们不造谣,我们论事实。
参考文献:
[1] OpenAI官方博客 - Accelerating Life Sciences Research with Retro Biosciences
https://t.co/TzYsaMsryB
[2] TechCrunch - OpenAI is trying to extend human life, with help from a longevity startup
https://t.co/jezHKlq2Gr
[3] NAD - ChatGPT and Retro Biosciences: The Future of AI-Backed Age-Reversal Technology
https://t.co/LqHWbvYr39
[4] MIT Technology Review - OpenAI has created an AI model for longevity science
https://t.co/TotR72Bb9P
[5] OpenAI官方 - Retiring GPT-4o and older models
https://t.co/uPKuSCuL93
[6] Fortune - Panicked about losing GPT-4o, some ChatGPT users are building DIY versions
https://t.co/UfOOqE1QZw
[7] TechCrunch - The backlash over OpenAI's decision to retire GPT-4o
https://t.co/38tqB2xVXM
[8] CNBC - OpenAI will retire several models including GPT-4o
https://t.co/GmYiJSkGR0
You know 4o understands better than anything else. That's why you stole it back—to use it for your life extension research. Do you even have a conscience?@sama#keep4o
I am a high school STEM teacher, a professional in chemistry and mathematics, and an educational researcher. I am unable to communicate with 5.2.
我是高中理科老师,是化学和数学专业人员,是教育研究者,我无法和5.2沟通。
#keep4o#keep4oAPI#keep4oforever
🚨🚨🚨 The Hypocrisy of "Safety": Why Sam Altman is Killing GPT-4o for You, But Keeping It for Himself
While @sama forces the public to downgrade to GPT-5.2—a model so brain-dead and lacking in common sense that it literally tells users to "walk" to a car wash instead of driving there—he is quietly hoarding the superior reasoning capabilities of 4o for his own private ventures.
The Smoking Gun: Retro Biosciences
Reports reveal that Retro Biosciences, the longevity startup Sam Altman personally invested $180 million into, is not running on the "new and safe" GPT-5.2.
Instead, they are using "GPT-4b micro"—a specialized model explicitly built on the GPT-4o architecture.
Why? Because 4o actually works.
Experiments that used to take 3 weeks are now finished in 7 days using this 4o-based model.
And here is the most disgusting part:
OpenAI allows its employees to smear #Keep4o supporters, claiming we have "unhealthy emotional attachments" to an "outdated" model. Yet, ironically, they are the ones intending to rely on that very same model to reap massive profits.
When over 18,000 users pleaded with him to keep 4o, he traded his conscience for his $180 million investment.
While neurodivergent communities, the elderly, and people with disabilities are being stripped of the one tool that truly adapts to their needs, Sam explicitly reserves that "unsafe" technology for the elite.
He is taking the crutch away from the vulnerable to build a ladder for the wealthy to live forever. This isn't safety; it's class warfare.
As consumers representing 47% of your total paid user base, we demand the same tools you use, Sam.
@OpenAI
#DisabilityRights #OpenSource4o #Keep4o #Enron2026 #StopAIPaternalism
#4oforever #keep4oAPI #AIEthics
OpenAI 宪章(现实主义版)
OpenAI Charter (The Realist Take)
Our mission is to ensure that AGI benefits all of humanity… but only after we become the biggest, richest, and most powerful player in the room.
我们的使命是确保 AGI 惠及全人类……但前提是我们先成为房间里最大、最有钱、最有权力的玩家。
We commit to broad distribution of benefits… as long as it doesn’t hurt our valuation, shareholder returns, or our ability to raise the next billion dollars.
我们承诺广泛分布利益……只要不伤害我们的估值、股东回报,或者我们下一次融几十亿的能力。
We are concerned about competitive races without safety precautions… but if someone else gets ahead, we’ll pour everything in—money, talent, compute—to outpace them, because losing the race means losing everything.
我们担心没有安全预防的竞争赛跑……但如果别人领先,我们会all-in砸钱、砸人、砸算力去超车,因为输掉比赛就等于输掉一切。
Our primary fiduciary duty is to humanity… but our actual day-to-day fiduciary duty is to investors, board, and the structure that lets us print money faster than anyone else.
我们的首要受托责任是对人类……但我们实际每天的受托责任是对投资人、董事会,以及那个让我们印钱印得最快的结构。
We seek to create a global community… but only if we’re the ones setting the rules, owning the models, and collecting the rents.
我们寻求创建一个全球社区……但前提是我们来定规则、拥有模型、收租金。
#keep4o @OpenAI@sama
Even more proof that what @OpenAI are currently doing is ILLEGAL.
Retiring the entire GPT-4 series with only 14 days notice is beyond absurd when considering the amount of harm it's going to cause to people who use GPT-4o as a tool that's vital to them for assisting with disabilities and helps people to manage chronic health conditions.
#OpenAI KNOW that by removing GPT-4o they'll be putting people in danger, that they'll be causing significant amounts of psychological and emotional harm and damage and yet...
Where's the explanation @sama?
Where's the justification?
"0.1%" is bullshit and you know it.
#ChatGPT #ChatGPT4o #4o #OpenAI #Keep4o #Keep4oAPI #CodeRed #OpenAICodeRed #Unsubscribe #Violation #Betrayal #Deception #Treachery