OpenAI
GPT-6.2 Astra
最大上下文
1.1M
最大输出
128K
发布日期
暂无
gpt-5.5GPT-5.5 被 OpenAI 描述为更智能的前沿模型,面向代码、研究、数据分析和专业知识工作。官方发布信息强调它在能力、速度和文档/结构化分析任务上的提升。它的描述应聚焦广泛的专业智能,而不是某一个单一细分场景。
上下文窗口
1.1M Token
最大输出
128K Token
发布日期
2026年4月23日
模态
| Token 阶梯 | 输入价格 | 输出价格 | 缓存读取 |
|---|---|---|---|
| <=272K | $5/M | $30/M | $0.5/M |
| >272K | $10/M | $45/M | $1/M |
推理
工具调用
Temperature 参数
附件
知识库
Endpoint 协议
GPT-5.5 结合高效推理、精确任务执行和大上下文视觉理解能力,面向复杂编程与专业工作。
在相近推理强度下使用更少的推理 token 获得良好结果,有利于长流程和多步骤任务。
在复杂工作中持续保持目标和约束,将产品意图转化为具体且可验证的结果。
可在 1,050,000 token 上下文中处理文本和图像,并保留细致视觉信息以分析文档和界面。
GPT-5.5 适合复杂软件工程、工具密集型 Agent 工作流,以及基于大规模资料回答问题的助手。
理解代码库、规划协同修改、实现功能,并使用可用开发工具验证结果。
在多步骤服务任务中选择并协调工具,持续遵守约束直至完成目标。
从大型文档集中检索相关资料、关联支持证据,并生成准确的用户答案。
import OpenAI from "openai"
const client = new OpenAI({
apiKey: process.env.TOKENHUB_API_KEY,
baseURL: "https://us-api.tokenhub.com/v1",
})
const result = await client.chat.completions.create({})
console.log(result.choices[0]?.message?.content)| 指数分数 | ||
|---|---|---|
| Artificial Analysis 综合能力指数 | Artificial Analysis 综合能力评分 | 54.8 |
| Artificial Analysis 代码能力指数 | Artificial Analysis 软件任务评分 | 59.1 |
| 知识与推理 | ||
| GPQA | 高阶科学问题求解 | 93.5% |
| HLE | 广泛专家级考试集合 | 44.3% |
| 代码与工程 | ||
| SciCode | 科学计算代码挑战 | 56.1% |
| Terminal-Bench Hard | 高难度终端任务执行 | 60.6% |
| 指令遵循与 Agent 任务 | ||
| IFBench | 指令约束遵循 | 75.9% |
| AA-LCR | 长上下文推理 | 74.3% |
| τ²-Bench | Agent 工作流任务 | 93.9% |
指标来源 Artificial Analysis
GPT-5.5 的能力、适用场景、限制和 TokenHub 使用指南。
GPT-5.5 是 OpenAI 面向前沿编程、推理和专业工作流的模型。
适合复杂编程、大量工具调用的自动化、面向客户的工作流,尤其适合重视专业输出质量的场景。
核心优势是强大的编程、专业推理和可靠工具调用。
它会使用更多计算,因此延迟和成本可能更高。 更看重吞吐量时,可考虑 GPT-5.4 Mini。
使用 TokenHub 显示的精确标识符,并按账户文档确认当前功能。
通过一个 API Key 接入 GPT-5.5 与更多 AI 模型。
媒体与讨论
精选与此模型相关的公开视频和帖子。
X (Twitter)
OpenAI ·
Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex. https://t.co/rPLTk99ZH5
https://pbs.twimg.com/amplify_video_thumb/2047374439407620096/img/hr_j95BZ5LxS1EJT.jpg查看原文OpenAI ·
GPT-5.5 delivers this step up in intelligence without compromising on speed. GPT-5.5 matches GPT-5.4 per-token latency in real-world serving, while performing better across nearly every evaluation we measured. It also uses significantly fewer tokens to complete the same Codex https://t.co/5mR46SM7mW
https://pbs.twimg.com/media/HGm8jVWbsAAwL60.png查看原文OpenAI ·
Update: GPT-5.5 and GPT-5.5 Pro are now available in the API.
查看原文Reddit
Introducing GPT-5.5 | OpenAI
Gerstlauer ·
查看原文Is GPT-5.5 actually a big step forward, or just a better efficiency story?
Single-Jack8 ·
展开全文
OpenAI saying GPT-5.5 can handle similarly hard tasks faster while using fewer tokens is interesting to me for one reason: that might matter more than a pure benchmark jump. A lot of model launches get framed as "smarter than the last one," but for people actually using these systems every day, I feel like the bigger question is: - does it produce better code with less fixing? - does research feel more reliable? - does long-context use break less often? - does lower token usage actually make heavy usage meaningfully cheaper? If the answer to those is yes, then GPT-5.5 could be a bigger release than it first looks. Not because it proves OpenAI is "winning," but because it moves the market toward models that are good enough *and* economically practical. That feels like a more important threshold than another small benchmark flex. Curious how people here see it: Is GPT-5.5 mainly a headline/positioning win for OpenAI, or do you think efficiency + speed is actually the next major frontier in the model race?
GPT-5.5 Instant is starting to roll out in ChatGPT.
Distinct_Fox_6358 ·
查看原文YouTube
Introducing GPT-5.5
OpenAI
查看原文OpenAI just dropped GPT-5.5... (WOAH)
Matthew Berman
查看原文GPT‑5.5 in 7 Minutes
Developers Digest
查看原文