GPT-5.5 发布了。它已在 OpenAI Codex 中可用，并正在向付费 ChatGPT 订阅用户推送。我提前获得了一些访问权限，发现这是一个快速、高效且能力极强的模型。和如今通常的情况一样，很难用语言具体描述它好在哪里——我让它构建东西，它就能精确地构建出我要求的东西！
本次发布有一个明显的遗漏——API：
API 部署需要不同的安全保障，我们正在与合作伙伴和客户密切合作，以满足大规模服务的安全要求。我们将很快在 API 中推出 GPT‑5.5 和 GPT‑5.5 Pro。
当我运行我的 pelican 基准测试时，我更喜欢使用 API，以避免 ChatGPT 或其他智能体框架中的隐藏系统提示词影响结果。
OpenClaw 后门
过去几个月，AI 领域持续存在的紧张点之一，是关于 OpenClaw 和 Pi 等智能体框架如何与大型提供商提供的 API 进行交互。
OpenAI 和 Anthropic 都提供广受欢迎的月度订阅服务，让用户能够以远低于原始 API 的价格访问其模型。
OpenClaw 直接集成了这一机制，随后被 Anthropic 阻止。这引发了一系列事件。OpenAI——最近聘请了 OpenClaw 的创建者 Peter Steinberger——看到了一个轻松赢得声誉的机会，并宣布欢迎 OpenClaw 继续通过其（开源）Codex CLI 工具所使用的相同机制，与 OpenAI 的订阅服务进行集成。
这是否意味着任何人都可以编写代码，通过集成 OpenAI 的 Codex 专用 API 来接入那些现有的订阅服务？
前几天，Jeremy Howard 问道：
有没有人知道 OpenAI 是否官方支持使用 Pi 和 Opencode（据我所知）所使用的 `/backend-api/codex/responses` 端点？
结果发现，OpenAI 的 Romain Huet 在 3 月 30 日发推文说：
我们希望人们能够随时随地使用 Codex 和他们的 ChatGPT 订阅！这意味着在应用程序中、在终端里，也可以在 JetBrains、Xcode、OpenCode、Pi 以及现在的 Claude Code 中使用。
这就是为什么 Codex CLI 和 Codex 应用服务器也是开源的原因！🙂
而 Peter Steinberger 回复 Jeremy 说：
OpenAI 订阅现已正式支持。
llm-openai-via-codex
于是……我让 Claude Code 对 openai/codex 仓库进行了逆向工程，弄清楚了身份验证令牌的存储方式，然后为我构建了 llm-openai-via-codex——这是一个用于 LLM 的新插件，它能获取你现有的 Codex 订阅，并用它来运行提示词！
（事后回想，我多希望当时用的是 GPT-5.4 或 GPT-5.5 预览版，那样会更有趣。我确实认真考虑过为了这个玩笑，用 Codex 和 GPT-5.5 从头重写整个项目，但最后还是决定不再在这上面花更多时间了！）
使用方法如下：
- 安装 Codex CLI，购买 OpenAI 套餐，登录 Codex
- 安装 LLM：`uv tool install llm`
- 安装新插件：`llm install llm-openai-via-codex`
- 开始提示：`llm -m openai-codex/gpt-5.5 '你的提示词写在这里'`
所有现有的 LLM 功能也应能正常使用——使用 `-a filepath.jpg/URL` 来附加图片，使用 `llm chat -m openai-codex/gpt-5.5` 开始持续对话，使用 `llm logs` 查看已记录的对话，以及使用 `llm --tool ...` 配合工具支持进行尝试。
再来点鹈鹕
我们来生成一只鹈鹕！
llm install llm-openai-via-codex
llm -m openai-codex/gpt-5.5 'Generate an SVG of a pelican riding a bicycle' 这是我得到的结果：
我觉得 GPT-5.4 的效果更好，所以我加上了 `-o reasoning_effort xhigh` 参数又试了一次：
那张图花了将近四分钟才生成，但我认为这次的效果好得多。
如果你比较一下 SVG 代码（默认版 vs xhigh 版），xhigh 版采用了截然不同的方法，CSS 的运用要多得多——正如那些渐变效果所展示的那样。xhigh 版使用了 9,322 个推理 token，而默认版只用了 39 个。
关于 GPT-5.5 的更多说明
GPT-5.5 最引人注目的一点是定价。一旦它在 API 中上线，其价格将是 GPT-5.4 的两倍——每 100 万输入 token 收费 5 美元，每 100 万输出 token 收费 30 美元，而 5.4 的价格分别是 2.5 美元和 15 美元。
GPT-5.5 Pro 的价格会更高：每 100 万输入 token 收费 30 美元，每 100 万输出 token 收费 180 美元。
GPT-5.4 将继续可用。考虑到 5.5 的价格是其两倍，感觉 5.4 之于 5.5，就像 Claude Sonnet 之于 Claude Opus。
Ethan Mollick 对 GPT-5.5 进行了详细评测，他让 GPT-5.5（以及 GPT-5.5 Pro）接受了一系列有趣的挑战。他的结论是：参差不齐的前沿依然存在——GPT-5.5 在某些方面表现出色，但在其他方面则表现不佳，且这种差异仍然难以预测。
2026 年 4 月 23 日
近期文章
- Kimi K3，以及我们仍能从鹈鹕基准测试中学到什么——2026 年 7 月 16 日
- 全新的 GPT-5.6 系列：Luna、Terra、Sol——2026 年 7 月 9 日
- sqlite-utils 4.0，现已支持数据库模式迁移——2026 年 7 月 7 日
这是通过半官方的 Codex 后门 API 为 GPT-5.5 准备的鹈鹕，作者 Simon Willison，发布于 2026 年 4 月 23 日。
chatgpt
llm
llm-reasoning
GPT-5.5 is out. It’s available in OpenAI Codex and is rolling out to paid ChatGPT subscribers. I’ve had some preview access and found it to be a fast, effective and highly capable model. As is usually the case these days, it’s hard to put into words what’s good about it—I ask it to build things and it builds exactly what I ask for!
There’s one notable omission from today’s release—the API:
API deployments require different safeguards and we are working closely with partners and customers on the safety and security requirements for serving it at scale. We’ll bring GPT‑5.5 and GPT‑5.5 Pro to the API very soon.
When I run my pelican benchmark I always prefer to use an API, to avoid hidden system prompts in ChatGPT or other agent harnesses from impacting the results.
The OpenClaw backdoor
One of the ongoing tension points in the AI world over the past few months has concerned how agent harnesses like OpenClaw and Pi interact with the APIs provided by the big providers.
Both OpenAI and Anthropic offer popular monthly subscriptions which provide access to their models at a significant discount to their raw API.
OpenClaw integrated directly with this mechanism, and was then blocked from doing so by Anthropic. This kicked off a whole thing. OpenAI—who recently hired OpenClaw creator Peter Steinberger—saw an opportunity for an easy karma win and announced that OpenClaw was welcome to continue integrating with OpenAI’s subscriptions via the same mechanism used by their (open source) Codex CLI tool.
Does this mean anyone can write code that integrates with OpenAI’s Codex-specific APIs to hook into those existing subscriptions?
The other day Jeremy Howard asked:
Anyone know whether OpenAI officially supports the use of the
/backend-api/codex/responsesendpoint that Pi and Opencode (IIUC) uses?
It turned out that on March 30th OpenAI’s Romain Huet had tweeted:
We want people to be able to use Codex, and their ChatGPT subscription, wherever they like! That means in the app, in the terminal, but also in JetBrains, Xcode, OpenCode, Pi, and now Claude Code.
That’s why Codex CLI and Codex app server are open source too! 🙂
And Peter Steinberger replied to Jeremy that:
OpenAI sub is officially supported.
llm-openai-via-codex
So... I had Claude Code reverse-engineer the openai/codex repo, figure out how authentication tokens were stored and build me llm-openai-via-codex, a new plugin for LLM which picks up your existing Codex subscription and uses it to run prompts!
(With hindsight I wish I’d used GPT-5.4 or the GPT-5.5 preview, it would have been funnier. I genuinely considered rewriting the project from scratch using Codex and GPT-5.5 for the sake of the joke, but decided not to spend any more time on this!)
Here’s how to use it:
- Install Codex CLI, buy an OpenAI plan, login to Codex
- Install LLM:
uv tool install llm - Install the new plugin:
llm install llm-openai-via-codex - Start prompting:
llm -m openai-codex/gpt-5.5 'Your prompt goes here'
All existing LLM features should also work—use -a filepath.jpg/URL to attach an image, llm chat -m openai-codex/gpt-5.5 to start an ongoing chat, llm logs to view logged conversations and llm --tool ... to try it out with tool support.
And some pelicans
Let’s generate a pelican!
llm install llm-openai-via-codex
llm -m openai-codex/gpt-5.5 'Generate an SVG of a pelican riding a bicycle' Here’s what I got back:
I’ve seen better from GPT-5.4, so I tagged on -o reasoning_effort xhigh and tried again:
That one took almost four minutes to generate, but I think it’s a much better effort.
If you compare the SVG code (default, xhigh) the xhigh one took a very different approach, which is much more CSS-heavy—as demonstrated by those gradients. xhigh used 9,322 reasoning tokens where the default used just 39.
A few more notes on GPT-5.5
One of the most notable things about GPT-5.5 is the pricing. Once it goes live in the API it’s going to be priced at twice the cost of GPT-5.4—$5 per 1M input tokens and $30 per 1M output tokens, where 5.4 is $2.5 and $15.
GPT-5.5 Pro will be even more: $30 per 1M input tokens and $180 per 1M output tokens.
GPT-5.4 will remain available. At half the price of 5.5 this feels like 5.4 is to 5.5 as Claude Sonnet is to Claude Opus.
Ethan Mollick has a detailed review of GPT-5.5 where he put it (and GPT-5.5 Pro) through an array of interesting challenges. His verdict: the jagged frontier continues to hold, with GPT-5.5 excellent at some things and challenged by others in a way that remains difficult to predict.
23rd April 2026
More recent articles
- Kimi K3, and what we can still learn from the pelican benchmark - 16th July 2026
- The new GPT-5.6 family: Luna, Terra, Sol - 9th July 2026
- sqlite-utils 4.0, now with database schema migrations - 7th July 2026
This is A pelican for GPT-5.5 via the semi-official Codex backdoor API by Simon Willison, posted on 23rd April 2026.
chatgpt
llm
llm-reasoning