DATA CUT 2026-07-27 116 活跃模型 680 A 类案例

模型档案

GPT-5.4 (xhigh)

GPT-5.4 (xhigh) 已有可核验 A 类案例,可优先进入选型对比。

已有真实案例

档案摘要 / Decision snapshot

先判断能不能用于选型。

证据完整度 100%
公开状态 已有真实案例

可进入对比

案例覆盖 5 条 A 类

有真实任务证据

主要适用 作为完整模型索引和厂商路线追踪入口

来自模型卡

谨慎场景 缺少真实案例时不要包装成推荐

边界优先

30 秒结论

可进入选型对比,但仍以案例证据为准。

来自公开模型资料库,并已补厂商/官方/模型家族证据;当前站内暂无可核验 A 类案例。

这页有来源、有风险说明、有公开可核验 A 类案例。

数据可信度 / 证据强度

已有真实案例

A 类案例

5

AA 评分

56.8

官方来源

2 个补充入口

案例来源

5 条可核验

来自公开资料与模型数据库 + Artificial Analysis + 2 个厂商/官方/模型家族证据入口;AA 不是唯一事实来源。

Evidence Distribution

案例证据分布

GPT-5.4 (xhigh) 当前关联 5 条 A 类案例;这里按任务、来源和复核时间观察证据结构。

A CASES 5 5 selected

Evidence Snapshot

证据快照覆盖

GPT-5.4 (xhigh) 的 5 条 A 类案例中,已有 5 条完成本地证据快照;0 条仍在快照队列。

SNAPSHOT 100%
已快照 5

可用 manifest 复核

待快照 0

等待落盘

部分快照 0

需补抓

需关注 0

需人工复核

证据完整度

5 / 5

模型卡 OK

Feishu Bitable

官方/厂商来源 OK

2 个入口

A 类案例 OK

5 条

公开产物 OK

已绑定

风险备注 OK

已标注

能力边界

用标签表达信号,不用标签替代证据。

查看证据方法
Coding Agent

通用/待核验

长上下文

暂无数据

研究任务

待核验

开放生态

平台/待核验

生命周期风险

厂商OpenAI
厂商页OpenAI
发布时间2026-03-05
模型 IDgpt-5-4-xhigh
上下文暂无数据
输出暂无数据
模态文本 / 多模态待核验
推理是 / 推理模型或 thinking 模式
价格官方未披露 / 暂无数据
可用平台OpenAI
AA 评分56.8
基础模型发布时间Aug 31, 2025
资料来源公开资料、厂商信息和案例库
A 类案例5

谱系位置

它在厂商路线中的位置

此页将 GPT-5.4 (xhigh) 放入 OpenAI 的当前 Atlas 路线中。若同厂前后代资料不足,先保留为路线追踪入口,不脑补谱系关系。

2026-03-05

GPT-5.4 (xhigh) 资料状态

来自公开资料与模型数据库 + Artificial Analysis + 2 个厂商/官方/模型家族证据入口;AA 不是唯一事实来源。

公开档案

模型状态

GPT-5.4 (xhigh) 当前标记为已有真实案例。

案例库

公开案例复核

已有 5 条可核验 A 类案例,达到完整补齐线

案例时间线

最近进入档案的 A 类案例

用采集时间呈现证据进入 Atlas 的顺序,方便复核来源新鲜度。

适合场景

  • - 作为完整模型索引和厂商路线追踪入口
  • - 推理/agentic workflow 候选

不适合

  • - 缺少真实案例时不要包装成推荐
  • - 缺字段时不要脑补价格、性能或上下文

真实案例

可核验 A 类案例精选

当前展示 5 条代表案例;排序优先看模型卡精选、证据可信度、快照状态和公开仓库/技术任务信号。案例总数仍为 5。

Representative Evidence

代表案例排序

当前第一代表案例是“KamakuraCrypto uses GPT-5.4 xhigh in a dual-Codex premortem review skill for Claude Code plans”。它会优先出现在本页,是因为该案例同时进入模型卡精选、具备 100/100 的证据可信度,并带有 已快照证据。

打开第一代表案例
排序池 5 条

模型卡精选优先

证据可信度 100/100

A+ 完整链路

快照状态 已快照

archive/evidence/gpt-5-4-xhigh-kamakuracrypto-claude-codex-review/manifest.json

技术信号 命中

Coding / repo / agent

KamakuraCrypto 使用 GPT-5.4 (xhigh) 处理代码审查和测试生成

KamakuraCrypto · GPT-5.4 (xhigh)

A
厂商:OpenAI 模型:GPT-5.4 (xhigh) 来源平台:github 最后复核:2026-06-27T09:20:44Z 证据快照:已快照 · 2 / 2 个证据目标 归档说明:已完成快照:2 / 2 个证据目标,可用本地 manifest 复核证据指纹。
证据可信度 100/100

A+ 完整链路 · 代码仓库证据

原始证据1 个公开产物复核通过代码仓库证据

KamakuraCrypto 公开的代码审查与测试案例,来源为 公开代码库,复核于 2026-06-27T09:20:44Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务:这是一个围绕代码审查和测试生成的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Claude Code skills that send a plan Claude just wrote to OpenAI's codex exec CLI for a structured premortem, using a dual-model setup that explicitly includes gpt-5.4 at xhigh reasoning alongside gpt-5.5.

公开产物:公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:The public repository provides installable /codex-review and /codex-review-full Claude Code skills, sandbox/full-access modes, convergence-loop instructions, blocker/question output schemas, and setup notes for reviewin…

模型作用:GPT-5.4 (xhigh) 在该案例中承担代码审查和测试生成相关的生成、分析、编排或实现角色。 原始资料写作:GPT-5.4 at xhigh reasoning is one of the two Codex reviewers in the premortem loop; it independently examines the proposed plan, reports blockers, likely failure modes, questions, confidence, and shippability status so …

A 类理由:有具体使用者、具体任务、公开原始证据和可访问产物,可公开核验。

风险边界:当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:Evidence is a public GitHub repository and README; it demonstrates a concrete reusable developer-tool artifact, though production usage outside the repository is not independently verified.

原始记录:KamakuraCrypto uses GPT-5.4 xhigh in a dual-Codex premortem review skill for Claude Code plans

已有真实案例 代码审查与测试公开代码库A 类可核验real_case auto_approved 进入模型卡精选

devdotbo 使用 GPT-5.4 (xhigh) 处理软件工程任务执行

devdotbo · GPT-5.4 (xhigh)

A
厂商:OpenAI 模型:GPT-5.4 (xhigh) 来源平台:github 最后复核:2026-06-27T08:59:16Z 证据快照:已快照 · 2 / 2 个证据目标 归档说明:已完成快照:2 / 2 个证据目标,可用本地 manifest 复核证据指纹。
证据可信度 100/100

A+ 完整链路 · 代码仓库证据

原始证据1 个公开产物复核通过代码仓库证据

devdotbo 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:59:16Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务:这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Claude Code plugin that delegates investigation, bug fixing, feature implementation, refactoring, read-only reviews, adversarial reviews, and dual Codex/Claude reviews to Codex jobs.

公开产物:公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:Public fork exposes slash commands such as /codex:execute, /codex:status, /codex:result, /codex:adversarial-review, and /codex:dual-review, with background job handling and stored final outputs.

模型作用:GPT-5.4 (xhigh) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:The plugin states execute tasks run in YOLO full-access mode with GPT 5.4 at xhigh reasoning; the common configuration section says the fork forces GPT 5.4 at xhigh reasoning for all execute tasks and uses it as the Cod…

A 类理由:有具体使用者、具体任务、公开原始证据和可访问产物,可公开核验。

风险边界:当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:Evidence is from the public repository README and repo page; deployment usage beyond the repository is not independently audited.

原始记录:devdotbo forces GPT 5.4 xhigh in a Claude Code Codex YOLO plugin for execution and dual review

已有真实案例 代码代理与软件工程公开代码库A 类可核验real_case auto_approved 进入模型卡精选

IvanLark 使用 GPT-5.4 (xhigh) 处理软件工程任务执行

IvanLark · GPT-5.4 (xhigh)

A
厂商:OpenAI 模型:GPT-5.4 (xhigh) 来源平台:github 最后复核:2026-06-27T08:59:16Z 证据快照:已快照 · 2 / 2 个证据目标 归档说明:已完成快照:2 / 2 个证据目标,可用本地 manifest 复核证据指纹。
证据可信度 100/100

A+ 完整链路 · 代码仓库证据

原始证据1 个公开产物复核通过代码仓库证据

IvanLark 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:59:16Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务:这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Non-invasive OpenClaw runtime shim that enables gpt-5.4 plus /think xhigh, hooks OpenClaw and bundled provider xhigh capability checks, and optionally injects priority service tier into the OpenAI Responses payload path.

公开产物:公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:Public repository provides a shim, installation/runtime instructions, compatibility notes for OpenClaw 2026.3.2, verification commands, rollback steps, and Chinese/English documentation.

模型作用:GPT-5.4 (xhigh) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:GPT-5.4 xhigh is the target model/effort combination being enabled; the shim preserves xhigh through OpenClaw capability checks and downstream provider handling so the configured agent can actually use that reasoning mo…

A 类理由:有具体使用者、具体任务、公开原始证据和可访问产物,可公开核验。

风险边界:当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:This is an integration/enablement artifact rather than an end-user application; README requires the user's backend/provider to already support gpt-5.4 with xhigh.

原始记录:IvanLark built an OpenClaw shim to keep gpt-5.4 xhigh enabled through provider capability checks

已有真实案例 代码代理与软件工程公开代码库A 类可核验real_case auto_approved 进入模型卡精选

noosi21 使用 GPT-5.4 (xhigh) 处理软件工程任务执行

noosi21 · GPT-5.4 (xhigh)

A
厂商:OpenAI 模型:GPT-5.4 (xhigh) 来源平台:github 最后复核:2026-06-27T08:59:16Z 证据快照:已快照 · 2 / 2 个证据目标 归档说明:已完成快照:2 / 2 个证据目标,可用本地 manifest 复核证据指纹。
证据可信度 100/100

A+ 完整链路 · 代码仓库证据

原始证据1 个公开产物复核通过代码仓库证据

noosi21 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:59:16Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务:这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Smart-contract audit CLI and Codex workstation for Solidity and DeFi protocols, combining skill-based audit prompts, parsing, symbolic/static tools, fuzzing, invariant checks, GitHub URL targets, and Foundry proof-of-co…

公开产物:公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:Public repository provides npm/Docker/Kali setup, codex-sol audit commands, 35 audit skills, GitHub target auditing, CI/diff audit options, and examples for checking reentrancy, flash loans, upgradability, and bug-bount…

模型作用:GPT-5.4 (xhigh) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:README binds the LLM path to GPT-5.4 xhigh: commands include --reasoning-effort xhigh, the Codex CLI integration is labeled GPT-5.4 xhigh Agent, and the agent reads AGENTS.md plus skill definitions before executing the …

A 类理由:有具体使用者、具体任务、公开原始证据和可访问产物,可公开核验。

风险边界:当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:Evidence is a public tool repository; specific vulnerability findings against third-party targets are examples rather than verified production audit reports.

原始记录:noosi21 uses GPT-5.4 xhigh in Codex Solidity for smart-contract audit workflows

已有真实案例 代码代理与软件工程公开代码库A 类可核验real_case auto_approved 进入模型卡精选

rwightman 使用 GPT-5.4 (xhigh) 处理软件工程任务执行

rwightman · GPT-5.4 (xhigh)

A
厂商:OpenAI 模型:GPT-5.4 (xhigh) 来源平台:github 最后复核:2026-06-27T08:59:16Z 证据快照:已快照 · 2 / 2 个证据目标 归档说明:已完成快照:2 / 2 个证据目标,可用本地 manifest 复核证据指纹。
证据可信度 100/100

A+ 完整链路 · 代码仓库证据

原始证据1 个公开产物复核通过代码仓库证据

rwightman 公开的代码代理与软件工程案例,来源为 公开代码库,复核于 2026-06-27T08:59:16Z。 Model Atlas 将它标记为 A 类证据,因为它同时具备具体使用者、具体任务、公开原始证据和可访问产物。

任务:这是一个围绕软件工程任务执行的真实任务,公开材料可以回溯到具体使用者和具体产物。 原始资料写作:Create a clean standalone PyTorch implementation of the Gemma 4 model family, including text, vision, and audio towers, generation, KV-cache, multimodal prompt expansion, and checkpoint conversion from Hugging Face safe…

公开产物:公开材料提供公开代码、README 或项目配置,可用于核验任务结果、项目形态和模型绑定关系。 原始资料写作:Public repository contains the gemma4-pytorch-codex package with working text generation, image preprocessing and image-conditioned generation, audio preprocessing, save/load, conversion utilities, tests, and documentat…

模型作用:GPT-5.4 (xhigh) 在该案例中承担软件工程任务执行相关的生成、分析、编排或实现角色。 原始资料写作:The GitHub repository title/description states the implementation was built using Codex with GPT-5.4 xhigh, tying the model to the code-generation and implementation task that produced the public artifact.

A 类理由:有具体使用者、具体任务、公开原始证据和可访问产物,可公开核验。

风险边界:当前判断基于公开材料;若产物下线、仓库变更或模型参与比例仅来自作者自述,需要在引用前重新复核。 原始资料写作:The exact GPT-5.4 xhigh evidence appears in the GitHub repo title/description rather than the README body; artifact contents are publicly accessible.

原始记录:rwightman produced a standalone Gemma 4 PyTorch implementation using Codex with GPT-5.4 xhigh

已有真实案例 代码代理与软件工程公开代码库A 类可核验real_case auto_approved 进入模型卡精选

数据缺口

只基于现有字段判断缺口;缺失项不会被猜测填充。

价格:待补公开资料 上下文:待补公开资料 输出:待补公开资料 官方/家族来源:已有 发布时间:已有 可用平台:已有 A 类案例:已有
  • - 补充方向:核验 价格 的官方/API/案例来源。
  • - 补充方向:核验 上下文 的官方/API/案例来源。
  • - 补充方向:核验 输出 的官方/API/案例来源。

下一步验证建议

  • - 继续保留 A 类案例的原始证据、产物页和版本快照。
  • - 补官方发布/API/System Card 来源;若官方未披露,继续标注“官方未披露”。
  • - 把 benchmark、教程、发布文保留为背景资料,不提升为真实案例。