OpenAI发布GPT-6.2 Astra与Sol模型及定价
GPT‑6.2 Astra and GPT‑6.2 Sol
GPT-6系列重大迭代,不仅公布全新旗舰与高效模型的详细定价策略,还包含第三方实测对比与典型缺陷复盘,对评估模型能力边界极具参考价值。
GPT‑6.2 Astra and GPT‑6.2 Sol
GPT‑6.2 Astra 和 GPT‑6.2 Sol
OpenAI:
OpenAI:
We’re introducing GPT‑6.2 Astra, our new frontier model for your hardest projects, and GPT‑6.2 Sol, our efficient, versatile model for everyday work.
我们推出 GPT‑6.2 Astra,这是专为最具挑战性项目打造的前沿模型;以及 GPT‑6.2 Sol,这是一款高效、多功能的模型,适用于日常工作任务。
They're available in the API as gpt-6.2-astra and gpt-6.2-sol. I shipped LLM 0.38 with support for the new models.
它们已在 API 中以 gpt-6.2-astra 和 gpt-6.2-sol 的形式提供。我发布了支持新模型的 LLM 0.38 版本。
6.2 Astra is 80% of the price of 6 Astra - $8/MTok input and $40/MTok output, the same price as Claude Fable 5.2 (which was also 80% of its predecessor). 6 Astra and Fable 5 were both $10/$50. The cached price is an even bigger discount - $0.40/million compared to $1/million for the older model. This is great news for longer running agentic sessions.
6.2 Astra 的价格是 6 Astra 的 80%——输入价格为每百万令牌 8 美元,输出价格为每百万令牌 40 美元,与 Claude Fable 5.2 的价格相同(后者也是其前代产品价格的 80%)。6 Astra 和 Fable 5 的价格均为 $10/$50。缓存价格折扣更大——旧模型为每百万令牌 1 美元,而新模型仅为每百万令牌 0.40 美元。这对于运行时间较长的智能体会话来说是个好消息。
GPT-6.2 Sol is the same $2/$10 price as 6.1 Sol and 6 Sol - which are half the price of GPT-5.6 Sol ($4/$20).
GPT-6.2 Sol 的价格与 6.1 Sol 和 6 Sol 相同,均为 $2/$10——这比 GPT-5.6 Sol($4/$20)便宜一半。
Here are the pelicans for GPT-6.2 Sol and for GPT-6.2 Astra. The 6.2 Astra max pelican looks like this:
以下是 GPT-6.2 Sol 和 GPT-6.2 Astra 的 pelicans(注:此处可能指某种可视化图表或对比图,保留英文原词)。6.2 Astra 的 max pelican 如下所示:
I have a full comparison grid here, but here's how the high/xhigh/max pelicans compare with the previous Sol and Astra releases:
我有完整的对比网格,但这里展示了 high/xhigh/max pelican 与之前发布的 Sol 和 Astra 版本的对比情况:
I also ran my most recent benchmark prompt against the two models, using "Ultra' effort levels in the Codex CLI:
我还使用 Codex CLI 中的“Ultra”努力级别,对我的最新基准测试提示词在这两个模型上进行了测试:
Make a three js animation where yonkers stands up as a big mech (cool stand up motion) and stomps around manhattan and queens and Brooklyn, use openstreetmap mapping data
制作一个 Three.js 动画,其中扬克斯(Yonkers)以酷炫的站立动作站起来,变成一台大型机甲,并在曼哈顿、皇后区和布鲁克林四处踩踏,请使用 OpenStreetMap 地图数据。
Here's Sol and here's Astra. Neither of them did a good enough job of morphing Yonkers into a big mech, at least in comparison to Claude Fable 5.2 and Gemini 4 Carbon.
这是 Sol 的结果,这是 Astra 的结果。与 Claude Fable 5.2 和 Gemini 4 Carbon 相比,它们都没有足够好地将扬克斯变形为一台大型机甲。
This video showing the Astra and Sol animations playing out at the same time:
这段视频同时展示了 Astra 和 Sol 动画的运行效果:
I created that video using GPT-6.2 Astra (Ultra) in the Codex desktop app by prompting:
我是通过在 Codex 桌面应用中提示以下内容,使用 GPT-6.2 Astra (Ultra) 创建该视频的:
https://static.simonwillison.net/static/2026/yonkers-gpt-6.2-astra-ultra/index.html and https://static.simonwillison.net/static/2026/yonkers-gpt-6.2-sol-ultra/index.html are two videos of the same thing by two different models
https://static.simonwillison.net/static/2026/yonkers-gpt-6.2-astra-ultra/index.html 和 https://static.simonwillison.net/static/2026/yonkers-gpt-6.2-sol-ultra/index.html 是两个不同模型制作的同一内容的视频
I need a video of the first 10 seconds where the top half of the video plays the first and the bottom half plays the second
我需要一段前 10 秒的视频,其中上半部分播放第一个视频,下半部分播放第二个视频
I need an overlaid label of "GPT-6.2 Astra" on the top half and "GPT-6.2 Sol" on the bottom half. These labels should be tastefully positioned.
我需要在画面上叠加标签:“GPT-6.2 Astra”位于上半部分,“GPT-6.2 Sol”位于下半部分。这些标签应美观地定位。
Make the video square and 2000x2000 pixels
将视频设为正方形,分辨率为 2000x2000 像素
Then later:
然后稍后:
OK make it 20s
好的,将其改为 20 秒
Here's the exported transcript.
以下是导出的转录文本。
The buzz around these new models is mixed at the moment. Astra 6.2 isn't scoring quite as high as Fable 5.2 on various benchmarks, and I've seen a few complaints that its communication style can be confusing. Theo Browne's review (in the form of a feedback letter he sent OpenAI during pre-release) has this useful example:
目前对这些新模型的讨论褒贬不一。Astra 6.2 在各种基准测试中的得分略低于 Fable 5.2,我也看到一些抱怨称其沟通风格令人困惑。Theo Browne 的评论(以他在预发布期间发给 OpenAI 的反馈信形式呈现)提供了一个有用的例子:
When this model is working on long tasks, it can pull in a lot of subagents and manage them pretty well. It will then proceed to randomly ask you some weird internal implementation detail bullshit that you could not under any earthly circumstance understand, much less have input on.
当该模型处理长任务时,它会调用大量子代理并很好地管理它们。随后,它会随机问你一些奇怪的内部实现细节废话,在任何情况下你都不可能理解,更不用说对此提出意见了。
I sent "/goal make a new level for Outer Wilds". 2 hours in, it's "blocked" because it wanted to ask me "does r24's reviewer write it's clips to 24_g521.mp4 or 24_p11.mp4?"
我发送了“/goal make a new level for Outer Wilds”。两小时后,它因受阻而停止,原因是它想问我“r24 的审阅者将其片段写入 24_g521.mp4 还是 24_p11.mp4?”
Tags: ai, openai, generative-ai, llms, llm-pricing, pelican-riding-a-bicycle, llm-release, theo-browne, gpt, yonkers-rising
标签:ai, openai, generative-ai, llms, llm-pricing, pelican-riding-a-bicycle, llm-release, theo-browne, gpt, yonkers-rising
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力