studioglobal
热门发现
报告已发布20 来源

GPT-5.5 Spud 核查：OpenAI 文档确认智能体编程，但未确认 Spud 发布

在本次核查的 OpenAI 公开文档中，未看到 GPT 5.5 Spud 作为已发布模型的官方页面；官方线索指向 GPT 5.4、GPT 5 Codex/Codex、Responses API、Agents SDK 和工具文档。 智能体编程和工具编排并不是传闻：Codex、函数调用、网页搜索、文件搜索、Shell、Apply Patch、Code Interpreter、Skills 等能力已有文档支撑。

使用 Studio Global AI 搜索并核查事实从“发现”浏览更多内容
17K0
AI-generated illustration of a fact-check comparing GPT-5.5 Spud rumors with documented OpenAI agent tools
GPT-5.5 Spud Fact-Check: Agentic Coding Is Documented, Spud Isn’tAgentic coding and tool orchestration are documented by OpenAI; GPT-5.5 Spud remains unconfirmed in the public docs reviewed.
AI 提示

Create a landscape editorial hero image for this Studio Global article: GPT-5.5 Spud Fact-Check: Agentic Coding Is Documented, Spud Isn’t. Article summary: In the public OpenAI docs reviewed here, GPT 5.5 Spud is not confirmed as a released model; the documented stack is GPT 5.4, GPT 5 Codex/Codex, the Responses API, Agents SDK, and official tools.. Topic tags: ai, openai, gpt 5, codex, ai agents. Reference image context from search candidates: Reference image 1: visual subject "[![Image 1: Lenny's Newsletter](https://substackcdn.com/image/fetch/$s_!8MSN!,w_40,h_40,c_fill,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaw" source context "🎙️ This week on How I AI: GPT 5.5, Claude Design, and GPT Images 2.0 hands-on reviews—plus an inside look at Memelord" Reference image 2: visual subject "[![Image 1: Lenny's Newsletter](https://substackcdn.com/image/f

openai.com

结论先说

关于 GPT-5.5 Spud 的争议，最容易混淆两件事：OpenAI 是否已经公开发布了这个名字的模型，以及 OpenAI 是否已经支持智能体式编程和依赖大量工具的编排。把两者分开看，证据很清楚：本次核查的 OpenAI 官方文档支持后者——GPT-5.4、GPT-5-Codex/Codex、Responses API、Agents SDK 和一系列工具都已被文档化；但这些文档没有确认 GPT-5.5 Spud 是一个公开发布的模型。[1][67][74][79][81][58][52]

逐条核查

说法结论有据可查的理解
OpenAI 已公开文档化一个已发布的 GPT-5.5 Spud 模型未证实本次核查的官方模型文档指向 GPT-5.4 和 GPT-5-Codex，而不是 GPT-5.5 Spud 的公开模型页。[1][67][74][79]
Spud 正以 GPT-5.5 传闻的形式流传有传闻依据TokenMix 明确把 GPT-5.5 预测标为推测，并称官方发布日期、模型卡和 API 定价尚未公布；India Today 也把 Spud 描述为传闻中的 GPT-5.5。[11][13]
OpenAI 支持智能体编程已确认OpenAI 称 GPT-5-Codex 针对智能体编程任务优化，代码生成指南也建议用 Codex 构建开箱即用的编码智能体。[79][81]
OpenAI 支持智能体编排已确认OpenAI 将智能体能力拆解为模型、工具、状态/记忆和编排等组件；Agents SDK 适用于应用自己掌控编排、工具执行、审批和状态的场景。[58][52]
依赖大量工具的工作流已有官方文档已确认OpenAI 文档覆盖函数调用、网页搜索、文件搜索、计算机使用、MCP/连接器、Shell、Apply Patch、Code Interpreter 和 Skills。[35][27][20][34][37][89][85][86][90]

未证实的部分：Spud 不是公开文档里的模型名

更稳妥的说法是：在本次核查的 OpenAI 公开文档中，GPT-5.5 Spud 尚未被确认为一个公开发布的模型。OpenAI 的模型文档把用户引向 GPT-5.4 作为最新模型指南；全部模型页面也把 GPT-5.4 描述为适用于智能体、编码和专业工作流的模型。[1][67][74]

这并不等于断言 Spud 不可能是内部代号。问题在于，现有证据不足以证明它已经成为公开模型发布：没有在这组资料中看到 GPT-5.5 Spud 的官方模型卡、API 端点、定价页面或 OpenAI 模型指南。Spud 相关证据主要来自第三方：TokenMix 称官方 GPT-5.5 发布日期、模型卡和 API 定价尚未公布；India Today 则把 Spud 表述为传闻中的 GPT-5.5。[11][13]

已确认的部分：GPT-5.4、GPT-5-Codex 与 Codex

在公开模型层面，本次资料的重心是 GPT-5.4 和 Codex。OpenAI 的全部模型页面列出 GPT-5.4，称其适合智能体、编码和专业工作流；GPT-5.4 模型页则把它称为面向复杂专业工作的前沿模型。[67][68]

在软件工程方向，GPT-5-Codex 是这组资料里最明确的命名编码模型。OpenAI 称 GPT-5-Codex 是为 Codex 或类似环境中的智能体编程任务优化的 GPT-5 版本，并且通过 Responses API 提供。[79] 代码生成指南还建议使用 Codex 构建开箱即用的编码智能体，并把 Codex 定位为 OpenAI 面向软件开发的编码智能体。[81]

智能体层：Responses API 与 Agents SDK

对开发者来说，关键不是追逐一个未经确认的模型名，而是看公开平台已经提供了什么。OpenAI 的智能体构建材料把平台描述为一组可组合的基础能力：模型、工具、状态/记忆和编排；同一材料还把 Responses API 描述为专门用于构建强大智能体的旗舰 API。[58]

如果应用需要自己掌控流程，Agents SDK 是更偏代码优先的路线。OpenAI 文档说明，当你的应用负责智能体编排、工具执行、审批和状态时，应使用 Agents SDK。[52] 另外，OpenAI 也有单独的编排与交接文档，说明多步骤或多智能体协作是已经文档化的平台模式，而不是必须依赖 Spud 这个未经确认的标签。[53]

工具层：调用函数、搜索网页、读取文件和修改代码

官方工具面相当宽。函数调用用于让模型调用开发者定义的工具；网页搜索则作为可通过 Responses API 使用的工具出现在文档中。[35][27]

除此之外，OpenAI 还文档化了文件搜索、计算机使用、MCP/连接器、Shell、Apply Patch、Code Interpreter 和 Skills。[20][34][37][89][85][86][90] 这些工具合在一起，支撑的是更接近真实开发和自动化场景的工作流：智能体可以检索私有内容、调用应用函数、搜索网络、连接外部系统、使用受控计算环境，并通过文档化接口应用代码变更。[20][35][27][37][89][85][86]

给开发者的实用判断

如果实际问题是：开发者现在能不能构建智能体编程和工具编排应用？从官方文档看，答案是可以。若问题变成：这些能力是否必须等 GPT-5.5 Spud？本次资料给出的答案是否定的。

  1. 先从已文档化的模型和指南入手，尤其是 GPT-5.4、GPT-5-Codex/Codex，而不是未经确认的模型名。[67][74][79][81]
  2. 当应用需要状态、工具执行、审批和流程编排时，优先查看 Responses API 和 Agents SDK。[58][52]
  3. 按工作流选择工具：函数调用用于自有 API，网页搜索用于实时网络检索，文件搜索用于私有内容，MCP/连接器用于外部系统，Shell、Apply Patch 或 Code Interpreter 则适用于需要代码执行或代码变更的场景。[35][27][20][37][89][85][86]
  4. 如果任务需要在不同专业步骤或不同智能体之间流转，再结合 OpenAI 的编排与交接指南。[53]

如何准确描述 Spud

目前最准确的表述是：GPT-5.5 Spud 未在本次核查的 OpenAI 公开文档中得到确认。不要把所谓发布日期、基准成绩、上下文窗口、API 可用性或定价当作这组资料已经证实的事实。TokenMix 明确表示官方 GPT-5.5 发布日期、模型卡和 API 定价尚未公布；India Today 也把 Spud 放在传闻中的 GPT-5.5 语境下描述。[11][13]

在 OpenAI 以这个名称发布官方文档之前，更有证据支撑的开发者路线仍然是：GPT-5.4、GPT-5-Codex/Codex、Responses API、Agents SDK，以及官方工具指南。[1][67][74][79][81][58][52]

Studio Global AI

Search, cite, and publish your own answer

Use this topic as a starting point for a fresh source-backed answer, then compare citations before you share it.

使用 Studio Global AI 搜索并核查事实

要点

  • 在本次核查的 OpenAI 公开文档中，未看到 GPT 5.5 Spud 作为已发布模型的官方页面；官方线索指向 GPT 5.4、GPT 5 Codex/Codex、Responses API、Agents SDK 和工具文档。
  • 智能体编程和工具编排并不是传闻：Codex、函数调用、网页搜索、文件搜索、Shell、Apply Patch、Code Interpreter、Skills 等能力已有文档支撑。
  • Spud 相关证据较弱：TokenMix 称官方发布日期、模型卡和 API 定价尚未公布，India Today 也将 Spud 描述为传闻中的 GPT 5.5。

人们还问

“GPT-5.5 Spud 核查：OpenAI 文档确认智能体编程，但未确认 Spud 发布”的简短答案是什么？

在本次核查的 OpenAI 公开文档中，未看到 GPT 5.5 Spud 作为已发布模型的官方页面；官方线索指向 GPT 5.4、GPT 5 Codex/Codex、Responses API、Agents SDK 和工具文档。

首先要验证的关键点是什么？

在本次核查的 OpenAI 公开文档中，未看到 GPT 5.5 Spud 作为已发布模型的官方页面；官方线索指向 GPT 5.4、GPT 5 Codex/Codex、Responses API、Agents SDK 和工具文档。 智能体编程和工具编排并不是传闻：Codex、函数调用、网页搜索、文件搜索、Shell、Apply Patch、Code Interpreter、Skills 等能力已有文档支撑。

接下来在实践中我应该做什么？

Spud 相关证据较弱：TokenMix 称官方发布日期、模型卡和 API 定价尚未公布，India Today 也将 Spud 描述为传闻中的 GPT 5.5。

接下来我应该探索哪个相关主题？

继续“Claude Opus 4.7、GPT-5.5、DeepSeek V4 与 Kimi K2.6：2026 基准对比与选型结论”以获得另一个角度和额外的引用。

打开相关页面

我应该将其与什么进行比较？

对照“DeepSeek V4 工程解析：1M 上下文、MoE 与 API 迁移要点”交叉检查此答案。

打开相关页面

继续你的研究

Comparativa de benchmarks 2026 entre Claude Opus 4.7, GPT-5.5, DeepSeek V4 y Kimi K2.6

Claude Opus 4.7、GPT-5.5、DeepSeek V4 与 Kimi K2.6：2026 基准对比与选型结论

Claude Opus 4.7、GPT-5.5、DeepSeek V4、Kimi K2.6：2026 基准测试怎么读

DeepSeek V4 工程架构示意图，包含 1M 上下文、MoE 专家路由和 API 服务化元素

DeepSeek V4 工程解析：1M 上下文、MoE 与 API 迁移要点

DeepSeek V4 工程解析：1M 上下文、MoE 与 API 迁移要点

木材市場與需求彈性概念的編輯插圖

Northwest vs. Southeast Timber 弹性题解：为什么是 larger; larger

Northwest 与 Southeast Timber：替代性越高，为什么答案是 larger; larger？

抽象 AI 晶片與多個聊天機器人平台競爭的示意圖，象徵 DeepSeek 挑戰 OpenAI、Claude、Gemini 與 Grok

DeepSeek会打败OpenAI、Claude、Gemini和Grok吗？关键不只在榜单

DeepSeek会打败OpenAI、Claude、Gemini和Grok吗？关键不只在榜单

研究对话

你

研究问题

Research and fact-check: GPT-5.5 Spud, Agentic coding and tool orchestration, including tool calling, web search, and tool-heavy workflows.

Studio Global AI40 来源

引用的答案

My fact-check is: OpenAI’s public docs strongly support agentic coding and tool orchestration capabilities, but I did not find a public OpenAI documentation trail here for a released model called “GPT-5.5 Spud.” In the surfaced official docs, the named public stack centers on GPT-5.4, GPT-5-Codex, the Responses API, Agents SDK, and multiple built-in tools; “Spud” appeared only in speculative third-party coverage in this check.[1][2][4][7][10][13]

Key findings

  • On the “GPT-5.5 Spud” claim, the official OpenAI docs surfaced here point to GPT-5.4 as the latest model guide and to GPT-5-Codex as a public coding model, not to a public “GPT-5.5” or “Spud” model page.[1][2]
  • The “Spud” material I found was third-party rumor coverage, and one of those pages explicitly says no official GPT-5.5 release date, model card, or API pricing has been announced and labels its forecast as speculation.[13]
  • OpenAI does publicly document agentic coding. The GPT-5-Codex page describes it as a version of GPT-5 optimized for agentic coding, and the code-generation guide says to use Codex for out-of-the-box coding agents.[2][3]
  • OpenAI also publicly documents orchestration. The Agents SDK guide says to use the SDK when your application owns orchestration, tool execution, approvals, and state, and OpenAI has a separate guide for orchestration and handoffs.[4][5]
  • Tool calling is officially supported through function calling.[6]
  • Web search is an official tool in the public docs, and the web-search guide shows it being used through the Responses API endpoint.[7][10]
  • Tool-heavy workflows are also supported through official tools including computer use, file search, MCP/connectors, shell, and apply patch, which together fit coding and automation workflows beyond plain text generation.[8][9][11][12][14]

Evidence notes

  • The official-doc evidence is much stronger than the “Spud” rumor coverage because it comes from OpenAI’s own developer documentation.[1][2][4][5][6][7][8][9][10][11][12][14]
  • So the reliable conclusion is not “GPT-5.5 Spud is confirmed,” but rather “OpenAI’s public platform already supports agentic coding and tool orchestration, while GPT-5.5 Spud remains unconfirmed in the public docs reviewed here.”[1][2][4][10][13]

Limitations / uncertainty

  • I did not directly verify any paywalled internal reporting in this pass.
  • If “Spud” is or was an internal codename, that is not confirmed by the public OpenAI docs surfaced here. Insufficient evidence.
  • Because this topic is release-sensitive, a future OpenAI announcement could change the answer quickly.

Summary

Fact-check verdict: “agentic coding and tool orchestration” is well-supported and officially documented by OpenAI.[2][4][5][6][7][8][9][10][11][12][14]
Fact-check verdict on “GPT-5.5 Spud”: not confirmed by the public official docs reviewed here; current evidence is rumor/speculation, not a public documented release.[1][2][13]

Sources

[1] OpenAI, “Using GPT-5.4” — https://developers.openai.com/api/docs/guides/latest-model
[2] OpenAI, “GPT-5-Codex Model” — https://developers.openai.com/api/docs/models/gpt-5-codex
[3] OpenAI, “Code generation” — https://developers.openai.com/api/docs/guides/code-generation
[4] OpenAI, “Agents SDK” — https://developers.openai.com/api/docs/guides/agents
[5] OpenAI, “Orchestration and handoffs” — https://developers.openai.com/api/docs/guides/agents/orchestration
[6] OpenAI, “Function calling” — https://developers.openai.com/api/docs/guides/function-calling
[7] OpenAI, “Web search” — https://developers.openai.com/api/docs/guides/tools-web-search
[8] OpenAI, “Computer use” — https://developers.openai.com/api/docs/guides/tools-computer-use
[9] OpenAI, “MCP and Connectors” — https://developers.openai.com/api/docs/guides/tools-connectors-mcp
[10] OpenAI, “Migrate to the Responses API” — https://developers.openai.com/api/docs/guides/migrate-to-responses
[11] OpenAI, “Shell” — https://developers.openai.com/api/docs/guides/tools-shell
[12] OpenAI, “Apply Patch” — https://developers.openai.com/api/docs/guides/tools-apply-patch
[13] TokenMix, “GPT-5.5 Release Date: 70% Odds for April, Spud Pretraining Done” — https://tokenmix.ai/blog/gpt-5-5-release-date-spud
[14] OpenAI, “File search” — https://developers.openai.com/api/docs/guides/tools-file-search

If you want, I can next turn this into a tighter “claim-by-claim verdict” table: confirmed, contradicted, or insufficient evidence.

来源

  • [1] Models | OpenAI APIdevelopers.openai.com

    Overview. Models. Latest: GPT-5.4. Text generation. Using tools. Overview. Models and providers. Running agents. [Evaluate agent…

  • [11] GPT-5.5 Release Date: 70% Odds for April, Spud Pretraining Donetokenmix.ai

    GPT-5.5 Release Date: 70% Odds for April, Spud Pretraining Done. GPT-5.5 Release Date: Spud Pretraining Done, What Developers Should Prepare For (2026). No official GPT-5.5 release date, no model card, no API pricing has been announced. Speculation Extrapol...

  • [13] OpenAI chief teases Spud model, says it is worth 2 years of research and major step toward AGI - India Todayindiatoday.in

    According to the OpenAI chief, Spud, which is rumoured to be GPT 5.5, is going to be much smarter than the company's previous AI models. So

  • [20] File search | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 const response = await openai.responses.create({ const response = await openai.responses.create({ model: "gpt-4.1", model: "gpt-4.1", input: "What is deep research by OpenAI?", input: "What is deep research by OpenAI?", tools: [{ tools:...

  • [27] Web search | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 curl " \ curl " \ -H "Content-Type: application/json" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $OPENAI API KEY" \ -H "Authorization: Bearer $OPENAI API KEY " \ -d '{ -d...

  • [34] Computer use | OpenAI APIdevelopers.openai.com

    The model can return visual observations by calling: display(base64 png string) """ """ from future import annotations from future import annotations import argparse import argparse import asyncio import asyncio import json import json import traceback impo...

  • [35] Function calling | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 from openai import OpenAI from openai import OpenAI client = OpenAI() tools = [{ "type": "function", "type": "function", "function": { "function": { "name": "get we...

  • [37] MCP and Connectors | OpenAI APIdevelopers.openai.com

    Using tools. MCP and Connectors. Computer use. File search. MCP servers. MCP and Connectors. Computer use. [File search](ht…

  • [52] Agents SDK | OpenAI APIdevelopers.openai.com

    Build agents in code with the OpenAI Agents SDK and grow into more advanced runtime patterns as needed. Use the Agents SDK pages when your application owns orchestration, tool execution, approvals, and state. Use Agent Builder only when you specifically wan...

  • [53] Orchestration and handoffs | OpenAI APIdevelopers.openai.com

    Text generation. Using tools. Overview. Quickstart. Agent definitions. Models and providers. Running agents. [Sandbox agents](

  • [58] Building agents - OpenAI Developersdevelopers.openai.com

    The OpenAI platform provides composable primitives to build agents: models , tools , state/memory , and orchestration . You can build powerful agentic experiences on our stack, with help in choosing the right models, augmenting your agents with tools, using...

  • [67] All models | OpenAI APIdevelopers.openai.com

    GPT-5.4. Best intelligence at scale for agentic, coding, and professional workflows ; gpt-5.4-pro. GPT-5.4 pro. Version of GPT-5.4 that produces smarter and more

  • [68] GPT-5.4 Model | OpenAI APIdevelopers.openai.com

    GPT-5.4 is our frontier model for complex professional work. Learn more in our latest model guide. Reasoning.effort supports: none (default), low, medium, high

  • [74] Using GPT-5.4 | OpenAI APIdevelopers.openai.com

    Latest: GPT-5.4. Using tools. Models and providers. Computer use. Reasoning models. Using realtime models. Latest: GPT-5.4. [Using tools](h…

  • [79] GPT-5-Codex Model | OpenAI APIdevelopers.openai.com

    Search the API docs. Realtime API. Model optimization. Legacy APIs. Using Codex. + Building frontend UIs with Codex and Figma. API. Building frontend UIs with Codex and Figma. A version of GPT-5 optimized for agentic coding in Codex. A version of GPT-5 opti...

  • [81] Code generation | OpenAI APIdevelopers.openai.com

    Learn how to use GPT-5.4 and Codex to generate code. This guide walks through your options for code generation with GPT-5.4 and Codex. Use Codex for out-of-the-box coding agents. Use Codex for out-of-the-box coding agents. Connect your codebase to Codex and...

  • [85] Apply Patch | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 from agents import Agent, ApplyPatchTool, Runner, apply diff from agents import Agent, ApplyPatchTool, R...

  • [86] Code Interpreter | OpenAI APIdevelopers.openai.com

    """ """ resp = client.responses.create( model="gpt-4.1", model="gpt-4.1", tools=[ { "type": "code interpreter", "type": "code interpreter", "container": {"type": "auto", "memory limit": "4g"} "container": {"type": "auto", "memory limit": "4g"} } ], instruct...

  • [89] Shell | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 curl -L ' \ curl -L ' \ -H "Authorization: Bearer $OPENAI API KEY" \ -H "Authorization: Bearer $OPENAI API KEY " \ -H "Content-Type: application/json" \ -H "Content-Type: application/json" \...

  • [90] Skills | OpenAI APIdevelopers.openai.com

    1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 curl -L ' \ curl -L ' \ -H "Content-Type: application/json" \ -H "Content-Type: application/json" \ -H "Authorization: Bearer $OPENAI API KEY" \ -H "Authorization: Bearer $OPENAI API KEY " \ -d '{ -d '{ "model...