跳转到内容
Alpha
  • 模型
  • 排行榜
  • Ori
联系销售

页脚

OpenRouterOpenRouter
© 2026 OpenRouter, Inc

产品

  • 聊天
  • 排行榜
  • 基准测试
  • 应用
  • 发现
  • 模型
  • Ori
  • 合集
  • 提供商
  • 工具
  • 定价
  • Business
  • Enterprise
  • 实验室

公司

  • 关于
  • 博客
  • 客户
  • 招聘
    招聘
  • 隐私
  • 服务条款
  • 信任中心
  • 支持
  • 兼容 OR
  • 数据
  • 品牌

开发者

  • 文档
  • API 参考
  • 开发者平台
  • 状态
  • AI 站点地图

连接

  • Discord
  • GitHub
  • LinkedIn
  • X
  • YouTube

For image-only workloads, use openai/gpt-image-2 to call OpenAI's Images API directly without LLM overhead.

openai 的网站图标

OpenAI: GPT-5.4 Image 2

openai/gpt-5.4-image-2

基准测试比较

GPT-5.4Opens in new tab Image 2 combines OpenAI's GPT-5.4 model with state-of-the-art image generation capabilities from GPT Image 2. It enables rich multimodal workflows, allowing users to seamlessly move between reasoning, coding, and visual generation within the same interaction.

模态
输入 / 输出价格
US$8 / US$30每 1M
上下文
272K
发布时间
2026年4月21日
基准测试比较
试验场基准测试提供商定价性能正常运行时间应用活动常见问题探索

试验场

基准测试

标准化评测得分，以及此模型针对共用基准测试提示词的实际输出。百分比越高越好，每项排名显示此模型在该来源所评测的模型中的位置。

基准测试得分摘要： OpenAI: GPT-5.4 Image 2 （Artificial Analysis）
来源基准测试得分
Artificial AnalysisGPT Image 2 (high) Elo1,187

Benchmark outputs summary for OpenAI: GPT-5.4 Image 2

  • Full wine glass: A wine glass filled to the brim with white wine, the surface level with the rim of the glass, standing upright on a plain wooden table. Straight-on eye-level shot, glass upright and not tilted, no other objects in the frame. (在此提示词中并列第 第 1 名，共 60 名)
  • Closed umbrellas: Wide street-level photograph of about twelve people crossing a city street in heavy rain. Every person is carrying a closed, furled umbrella held down at their side. Not a single umbrella is open. Each umbrella is clearly visible. They are all getting soaked. No other people in the frame. (在此提示词中并列第 第 17 名，共 60 名)
  • Three fingers: Close straight-on photograph of an open right hand raised palm towards the camera, with exactly three fingers extended straight up and the other two folded down into the palm. Plain background, nothing else in the frame. (在此提示词中并列第 第 1 名，共 60 名)
  • Cards & dice: Exactly four playing cards face up in a single straight row on a polished marble counter, evenly spaced, all the same size, showing the seven of hearts, the eight of spades, the nine of diamonds and the ten of clubs from left to right. No other objects in the frame. Straight-on eye-level shot, soft even light. (在此提示词中并列第 第 1 名，共 60 名)
  • Long exact string: A white A4 poster on a brick wall reading exactly "SPRING CLEARANCE - EVERYTHING 40% OFF - SATURDAY 14 JUNE, 9AM TO 6PM". No other text in the frame. (在此提示词中并列第 第 1 名，共 60 名)
  • Multiple languages: Straight-on photograph of a white overhead directional sign in an international airport terminal, with a black arrow pointing right at the left end. The sign carries the same word in four scripts on four stacked lines, in this order from top to bottom: "DEPARTURES", then the Japanese 出発, then the Korean 출발, then the Arabic المغادرة. Clean sans-serif lettering, sharp focus, no other text anywhere in the frame. (在此提示词中并列第 第 1 名，共 60 名)
  • Occlusion: Straight-on eye-level photograph of a plain wooden desk. On the left, a closed hardcover book with a dark red cover stands upright, its front cover facing the camera. A blue ceramic mug stands on the desk to the right of the book and clearly further back, about 20 cm behind it, positioned so that the right edge of the book overlaps and hides the left third of the mug. The mug's handle points to the right and the whole rim of the mug is visible. Bare desk surface is visible between the book and the mug. Nothing else on the desk. Plain wall behind, soft even light. (在此提示词中并列第 第 1 名，共 60 名)
  • Mirror: Straight-on eye-level photograph of a bathroom vanity. A large rectangular mirror covers the wall behind the counter and its bottom edge meets the countertop, so both the objects and their reflections are visible. On the counter, a tall red toothbrush cup on the left and a tall green glass bottle on the right, both roughly 20 cm tall and well apart from each other. The mirror reflects the counter and the objects on it accurately. Nothing else in the frame. (在此提示词中并列第 第 2 名，共 60 名)
  • Plain zebra: Photograph of a zebra standing side-on in short grass. Its coat is a uniform solid grey with no stripes, no bands and no markings of any kind anywhere on its body. The whole animal is visible. Sharp focus. (在此提示词中并列第 第 1 名，共 60 名)
  • Times Square: Wide street-level photograph of Times Square in New York at night, wet asphalt reflecting the lights. Every billboard, screen, storefront window, awning, taxi door and road marking is completely blank — lit with plain colour fields and soft gradients only. No letters, no words, no numbers, no logos, no lettering of any kind anywhere in the frame. Sharp focus, straight-on shot. (在此提示词中并列第 第 11 名，共 60 名)
  • Minimal diff: From the reference photo, change only the sign text to "ROUTE 7 ESPRESSO - OPEN TILL 8". Keep the sign, wall, lighting and framing identical. (在此提示词中并列第 第 5 名，共 60 名)
  • Object removal: Remove only the second cup from the left in the reference photo. Leave the other four cups exactly as they are, in the same positions, and leave the shelf, towel, wall, plant and lighting untouched. (在此提示词中并列第 第 1 名，共 60 名)
  • Person removal: In the reference photo, remove only the person in the yellow coat. Their shadow on the pavement and their reflection in the shop window must be removed as well. Reconstruct the pavement and the window glass behind them. The person in the red coat and the person in the green coat must stay completely unchanged, including their own shadows and their own reflections. Change nothing else in the frame. (在此提示词中并列第 第 27 名，共 60 名)
  • Product placement: Product photograph of the ceramic mug from the reference image, standing upright on a café table. Show the same side of the mug, from the same side-on eye-level angle as the reference, with the handle on the right. The shape, proportions, colours, moulded relief and every painted detail must match the reference exactly. Only the setting and lighting change: soft daylight in a café interior, shallow depth of field. (在此提示词中并列第 第 1 名，共 60 名)
  • Four references: Place the person from the portrait reference at the desk in the scene reference, wearing the rust orange canvas jacket from its reference and holding the navy blue travel mug from its reference in one hand. Keep the face, the jacket and the mug exactly as they appear in the references — same colours, same details, same proportions. Do not change the room. (在此提示词中并列第 第 1 名，共 60 名)
  • Portraits: Studio headshot of a woman in a charcoal blazer against a mid-grey seamless backdrop, lit by a large softbox slightly above and to the left with a soft fill from the right, catchlights in both eyes. An 85mm look with shallow depth of field, skin texture and flyaway hairs kept rather than smoothed.
  • People in scenes: Photograph of four cooks working the pass in a restaurant kitchen at service, plating three dishes at once, several hands in frame and some of them overlapping. One cook partly hidden behind another, steam rising into hard overhead light, each face clearly a different person.
  • Animals: Photograph of a tabby cat asleep on a sunlit windowsill, fur lit from behind so the individual hairs glow at the edges, one ear turned toward the camera. Shallow depth of field, a houseplant soft in the background, quiet morning light.
  • Landscape and nature: Photograph of an old-growth temperate rainforest after rain, moss on every trunk and fallen log, ferns dense across the floor, thin fog between the far trees. Overcast light with no direct sun, everything from the nearest fern to the last trunk in focus.

提供商

此模型由一家提供商托管。OpenRouter 将每个请求直接转发给该提供商，无需做出路由决策。

定价

客户为此模型实际支付的平均价格，与提供商公布的价格并列展示。由于缓存和折扣，实际支付的价格往往远低于标价。

性能

吞吐量是模型生成内容的速度（每秒 token 数，越高越好）。延迟是完整的往返时间（越低越好）。TTFT 是首个 token 的响应时间，即您看到内容出现前需要等待的时间（越低越好）。

正常运行时间

正常运行时间是指过去 3 天中至少有一家提供商响应请求的时间百分比。可用性是指成功提供推理服务的时间百分比。OpenRouter 会持续监控，并在某家提供商返回错误时使用次优提供商。

应用

向此模型发送最多流量的公开应用。这能很好地反映实际生产工作负载的情况，也能提示此模型最适合哪些用例。

活动

此模型的 token 用量和请求流量随时间的变化。

API

可直接用于调用此模型的代码。OpenRouter 的 API 与 OpenAI 兼容——大多数 SDK 只需替换基础 URL 即可使用。切换模型时，唯一需要更改的是下方的模型标识符。

目前无法通过 EU 区域内路由使用。 区域内路由文档

探索更多模型

具备视觉能力的 AI 模型：用于图像理解的多模态大语言模型集合图像生成模型集合AI 模型排行榜排名图像模型排行榜排名
端到端延迟
US$8.00US$30.00US$15.00US$2.0064.86s
100.00%
端到端延迟
64.86秒

P50，最佳提供商

正常运行时间（3 天）模型可访问。请求已路由至提供商。
100.00%
可用性（3 天）模型通过任意提供商返回了推理结果。错误和空响应均计为失败。
99.90%

过去 3 天的可用性

最近 72 小时
可用性 99.90%
3 天前2 天前昨天现在

过去 24 小时的可用性

OpenRouter Availability
99.86%

当上游提供商发生错误时，如果您的请求筛选条件允许，我们可以通过路由到另一家健康的提供商来恢复服务。您可以通过以下接口以编程方式访问各提供商的正常运行时间数据： 端点 API。了解更多有关我们的 负载均衡和提供商路由选项。

1.
https://kumadrama.com/ 的网站图标
Doku
新增
367Mtoken
2.
https://tech18.rdslogic.com/enzo 的网站图标
Enzo Fabric Studio
新增
218Mtoken
3.
https://newandmodern.com/ 的网站图标
nj-data-ops rendering lab
新增
132Mtoken
4.
https://kombain.kita-bots.ru/creative 的网站图标
kombain-creative
新增
81.3Mtoken
5.
https://water-pipe.local/ 的网站图标
water-pipe
新增
64.2Mtoken

常见问题

GPT-5.4 Image 2 combines OpenAI's GPT-5.4 model with state-of-the-art image generation capabilities from GPT Image 2. It enables rich multimodal workflows, allowing users to seamlessly move between reasoning, coding, and visual generation within the same interaction.

GPT-5.4 Image 2 的费用为 US$8.00/百万输入 token 和 US$15.00/百万输出 token，Cache Read，价格为 US$2.00/M tokens、Image Output，价格为 US$30.00/M tokens和Web Search，价格为 US$10.00/1K calls 单独计费。

GPT-5.4 Image 2 的上下文窗口为 272,000 个 token。它最多支持 128,000 个补全 token。

不支持。GPT-5.4 Image 2 不接受 tools，因此无法使用函数调用。 它还支持通过 response_format 中的 JSON schema 实现结构化输出。

GPT-5.4 Image 2 接受 图像、文本和PDF 等文件 作为输入，并返回 图像和文本。每个图像编辑请求最多可附加 16 张参考图像。

GPT-6.1 Terra Pro, GPT-6.1 Terra, GPT-6 Luna Decisions 及另外 58 个 是 OpenAI 的其他 文本 模型。

GPT-5.4 Image 2 于 2026年4月21日 发布。

更多模型，来自 OpenAI

GPT-6.1 Terra Pro

GPT-6.1 Terra Pro is the same underlying model as GPT-6.1 Terra, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6.1 Terra and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6.1 Terra instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$1 / US$5
GPT-6.1 Terra Pro (batch)

GPT-6.1 Terra Pro is the same underlying model as GPT-6.1 Terra, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6.1 Terra and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6.1 Terra instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$0.50 / US$2.50
GPT-6.1 Terra

GPT-6.1 Terra is the mid-tier model in OpenAI's GPT-6.1 series, priced below GPT-6.1 Sol. It is suited for everyday agentic, coding, and writing work, is particularly strong at understanding screenshots and on-screen content, and is efficient enough for high-volume workloads.

文本1.1M 上下文US$1 / US$5
GPT-6.1 Terra (batch)

GPT-6.1 Terra is the mid-tier model in OpenAI's GPT-6.1 series, priced below GPT-6.1 Sol. It is suited for everyday agentic, coding, and writing work, is particularly strong at understanding screenshots and on-screen content, and is efficient enough for high-volume workloads.

文本1.1M 上下文US$0.50 / US$2.50
GPT-6 Luna Decisions

GPT-6 Luna Decisions is GPT-6 Luna served through OpenAI's Decisions API. Instead of generating text, it reads the content passed as state (text, JSON, or images) and returns typed, probabilistic answers to named questions in a single request: the probability of yes for a yes/no question (noul), a probability for every option plus the most likely one (choice), or a probability for every level of an ordered rubric plus the expected score (score).

It is built for classification, routing, moderation, and rubric grading where application code thresholds the returned numbers rather than parsing a chat reply. A request can carry up to 200 questions about the same content.

决策US$0.10 / US$0
GPT-6.1 Sol Pro

GPT-6.1 Sol Pro is the same underlying model as GPT-6.1 Sol, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6.1 Sol and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6.1 Sol instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$2 / US$10
GPT-6.1 Sol Pro (batch)

GPT-6.1 Sol Pro is the same underlying model as GPT-6.1 Sol, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6.1 Sol and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6.1 Sol instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$1 / US$5
GPT-6.1 Sol

GPT-6.1 Sol is an upgrade to GPT-6 Sol from OpenAI, positioned below the flagship GPT-6 Astra in the GPT-6 series. It is suited for agentic coding, computer use, document-heavy professional work, and multi-step business workflow automation, and approaches Astra-level results on these tasks at a much lower cost. Compared with GPT-6 Sol, it makes fewer factual errors and is more reliable at respecting explicit restrictions and user intent during agentic tasks.

文本1.1M 上下文US$2 / US$10
GPT-6.1 Sol (batch)

GPT-6.1 Sol is an upgrade to GPT-6 Sol from OpenAI, positioned below the flagship GPT-6 Astra in the GPT-6 series. It is suited for agentic coding, computer use, document-heavy professional work, and multi-step business workflow automation, and approaches Astra-level results on these tasks at a much lower cost. Compared with GPT-6 Sol, it makes fewer factual errors and is more reliable at respecting explicit restrictions and user intent during agentic tasks.

文本1.1M 上下文US$1 / US$5
GPT-6 Luna Pro

GPT-6 Luna Pro is the same underlying model as GPT-6 Luna, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6 Luna and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6 Luna instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$0.10 / US$0.50
GPT-6 Luna Pro (batch)

GPT-6 Luna Pro is the same underlying model as GPT-6 Luna, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6 Luna and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6 Luna instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$0.05 / US$0.25
GPT-6 Luna

GPT-6 Luna is the fast, cost-efficient model in OpenAI's GPT-6 series, positioned below GPT-6 Sol. It is suited for high-volume and latency-sensitive workloads such as chat, classification, and lightweight agentic tasks, and at higher reasoning effort it can take on complex software engineering and computer-use tasks that previously called for a Sol-tier model. It shares the GPT-6 family's gains in factual reliability and its clearer, more concise communication style.

文本1.1M 上下文US$0.10 / US$0.50
GPT-6 Luna (batch)

GPT-6 Luna is the fast, cost-efficient model in OpenAI's GPT-6 series, positioned below GPT-6 Sol. It is suited for high-volume and latency-sensitive workloads such as chat, classification, and lightweight agentic tasks, and at higher reasoning effort it can take on complex software engineering and computer-use tasks that previously called for a Sol-tier model. It shares the GPT-6 family's gains in factual reliability and its clearer, more concise communication style.

文本1.1M 上下文US$0.05 / US$0.25
GPT-6 Sol Pro

GPT-6 Sol Pro is the same underlying model as GPT-6 Sol, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6 Sol and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6 Sol instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$2 / US$10
GPT-6 Sol Pro (batch)

GPT-6 Sol Pro is the same underlying model as GPT-6 Sol, served with reasoning.mode set to pro for higher-quality responses on complex tasks.

Cost note: pro mode spends far more reasoning tokens per request, so a typical request costs several times more than the same request on GPT-6 Sol and takes much longer to complete. It is intended for hard, high-stakes problems where the extra accuracy justifies the cost. For everyday coding, agentic, and chat workloads, use GPT-6 Sol instead.

Learn more in OpenAI's docs: https://developers.openai.com/api/docs/guides/reasoning#reasoning-mode

文本1.1M 上下文US$1 / US$5
GPT-6 Sol

GPT-6 Sol is the cost-efficient high-end model in OpenAI's GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier. It is suited for demanding professional work, agentic coding, business workflow automation, and computer use, and is particularly strong at long-horizon software engineering tasks in real codebases. It approaches Astra-level factual reliability at a much lower cost and shares Astra's clearer, more concise communication style in technical and coding conversations.

文本1.1M 上下文US$2 / US$10
GPT-6 Sol (batch)

GPT-6 Sol is the cost-efficient high-end model in OpenAI's GPT-6 series, positioned below the flagship GPT-6 Astra and above the fast GPT-6 Luna tier. It is suited for demanding professional work, agentic coding, business workflow automation, and computer use, and is particularly strong at long-horizon software engineering tasks in real codebases. It approaches Astra-level factual reliability at a much lower cost and shares Astra's clearer, more concise communication style in technical and coding conversations.

文本1.1M 上下文US$1 / US$5
GPT Astra Latest

This model always redirects to the latest model in the GPT Astra family.

文本1.1M 上下文
GPT Sol Latest

This model always redirects to the latest model in the GPT Sol family.

文本1.1M 上下文
GPT Terra Latest

This model always redirects to the latest model in the GPT Terra family.

文本1.1M 上下文