OpenAI reportedly delays GPT-6.2 and GPT-6.5 over safety and alignment issues
A leaked post and a tipster claim OpenAI is delaying GPT-6.2 and GPT-6.5 releases due to safety and alignment concerns, with delays lasting several weeks. The tipster notes OpenAI faces challenges turning over 10 trillion parameter pretraining results into usable products, and community discussion has surged on Twitter.
Coverage timeline
量子位量子位
GPT-6.2难产!OpenAI怕了 henry 2026-11-18 15:53:29 来源: 量子位 OpenAI也终于活成了2022年谷歌的样子 henry发自 凹非寺 量子位 | 公众号QbitAI OpenAI也终于活成了2022年谷歌的样子: 预告和承诺一波接一波,新模型却迟迟不见踪影,GPT-6.2更是一拖再拖。 而背后的核心原因,又是“安全”。 一则流传甚广的“已删”贴文中,OpenAI工程师Joe爆料: OpenAI已经成了全世界最把安全当回事的公司之一。 OpenAI现在是一家安全公司。 那意思就是,在OpenAI,安全焦虑已经被排到了首位,发布新模型反而成了次要任务。 该说不说,这多少有点谷歌味了。 (2022年的谷歌,手握领先技术,却因为安全和品牌声誉顾虑迟迟不敢发布产品,眼睁睁看着OpenAI抢走了生成式AI的先发优势。) 而就在刚刚,AI爆料博主Token Gremlin又给出了更具体的说法:OpenAI的新一代模型正因 安全与对齐问题 推迟发布,延期已经持续数周。 而且,不只是按发布顺序接棒的GPT-6.2,他还同时点名了GPT-6.5,并指出OpenAI正面临一个棘手的问题: 如何把超过10万亿参数的预训练成果,变成公众真正能用上的产品。 目前,这一爆料已经引发社区热议,相关话题也登上了推特热榜。 说起来,这事之所以值得关注,是因为早在11月初,社区里就陆续有人声称,自己已经被灰度到了GPT-6.2。 就连The Information也曾在11月4号透露,OpenAI的新Sol、Astra和cyber模型最快可能在当周五,也就是11月6日推出。 结果,眼瞅着半个月过去了,不仅新Astra还没等到,其他型号也是一鸽再鸽。 反观好兄弟Anthropic,倒是在11月9号接连发布Claude Fable 5.2和Mythos 5.2~ 难不成,OpenAI真就在安全问题上画地为牢了,厉娜! 还是说。。。奥特曼准备等到19日A社上市,再找机会狙击一手? GPT-6.2还在路上 按照Token Gremlin最新的说法,奥特曼送给达里奥的上市大礼包,恐怕要因为安全对齐问题而延期了。 除了延期数周,他还给出了一个更悲观的判断:这些问题短时间内很难得到解决,模型发布节奏也难以迅速恢复。 即使OpenAI近期推出GPT-6.2,也可能不得不带着大量
Related stories
Unconfirmed reports say OpenAI's GPT-6.2 Astra and Sol delayed over safety and alignment issues
Developers had expected OpenAI's Astra and Sol models around November 6 after the identifiers gpt-6.2-astra and gpt-6.2-sol appeared in Codex usage records and on a Google Cloud-related page, but no launch occurred. On November 8, social media users circulated unconfirmed claims that OpenAI found serious safety and alignment problems late in the process requiring emergency fixes, possibly taking several days; OpenAI has not confirmed the release plan or any safety-related delay. The discussion traces back to a November 5 statement by third-party evaluator Vals AI about models from frontier labs.
1 reports · 1 sources

OpenAI launches GPT-6.2 Astra and Sol, cuts Astra API prices 20%
OpenAI introduced GPT-6.2 Astra, a frontier model for complex professional work, and GPT-6.2 Sol, an efficient model for everyday tasks, available in ChatGPT Work, Codex, and the API. Astra's API pricing is $8 per million input tokens and $40 per million output tokens, a 20% cut from GPT-6 Astra, while Sol retains its predecessor's pricing. Benchmarks show Astra trailing only Claude Fable 5.2 in one index and tied in another, though a critical review cites usability frustrations and poor long-task performance.
7 reports · 4 sources

OpenAI GPT-6.2 Astra Spotted in Codex Usage Stats, Official Listing Absent
User screenshots and reports indicate OpenAI may be conducting a wide gray-area rollout of a model internally labeled GPT-6.2 Astra, appearing in Codex usage statistics while the official model catalog still lacks it. The model may skip the reportedly canceled GPT-6.1 Astra, with community speculation suggesting fixes to prior safety issues and a version bump. Reports claim performance gains over GPT-6 Astra and lower costs, but official confirmation is pending.
1 reports · 1 sources

OpenAI releases GPT-6.1 Terra with computer-use focus, API price halved
OpenAI announced GPT-6.1 Terra, a fast, efficient model emphasizing visual reasoning and computer use, priced at $1 per million input tokens and $5 per million output tokens. The standard API price is 50% lower than GPT-5.6 Terra and GPT-6.1 Sol, and the model is available in ChatGPT Work, Codex, and via API (gpt-6.1-terra) for Plus, Pro, Enterprise, and Edu users. In the Agents' Last Exam benchmark, Terra scored 15 points higher than GPT-5.6 Terra at medium reasoning intensity, with about one-third the cost, and matched GPT-6 Astra's performance at one-tenth the token price.
4 reports · 4 sources

News updates every hour. Subscribe in your RSS reader to get new stories automatically.
Subscribe