GPT-Image-2.6 Flare

GPT-Image-2.6 Flare is an image generation and editing model designed for rapid iteration. It can generate or edit images based on text and image inputs and is suitable for interactive image generation and editing workflows. The documentation mentions complexity settings, markup-guided editing, SVG output, and animated GIF output. ### Pricing Billed per million tokens, in USD: - Text input: 5.00 - Cached text input: 1.25 - Image input: 8.00 - Cached image input: 2.00 - Image output: 30.00 Supports only text and image inputs and image output; audio and video are not supported.

GPT-Image-2.6 Sunburst

GPT-Image-2.6 Sunburst supports image generation and editing based on text and image inputs, making it suitable for precise composition and image editing workflows. The documentation mentions complexity settings, markup-guided editing, SVG output, and animated GIF output. ### Input and Output - Input: text, images - Output: images - Audio and video modalities are not supported ### Official Pricing All prices below are in USD per 1 million tokens: | Type | Input | Cached Input | Output | | --- | ---: | ---: | ---: | | Text Tokens | 5.00 | 1.25 | — | | Image Tokens | 8.00 | 2.00 | 30.00 |

GPT Image 2

GPT Image 2 是我们最先进的图像生成模型，用于快速、高质量的图像生成和编辑。它支持灵活的图像尺寸和高保真图像输入。

Nano Banana

Nano Banana 是 Google Gemini 的模型。 模型族：gemini-flash。 知识截止：2025-06。 模态：text、image → text、image。 上下文长度：32,768 tokens。

GPT-Image-2.5 Flare

GPT-Image-2.5 Flare 是面向日常高质量图像生成的高速模型，支持文本和图像输入，仅输出图像。 支持 low、medium、high、xhigh、max 和 auto 六种质量设置，可通过 Image API 直接调用，也可作为 Responses API 图像生成工具的模型使用。支持图像生成、图像编辑和 Batch 处理，不支持音频或视频。 ### Token 价格 以下金额均为 USD / 100 万 Token： - 文本输入：5；缓存文本输入：1.25。 - 图像输入：8；缓存图像输入：2。 - 图像输出：标准处理 30；Batch 处理 15。 模型不输出文本，因此没有文本输出计费。Token 单价与 GPT Image 2 相同，但 GPT Image 2 的计算器不能用于估算 GPT Image 2.5 的 Token 消耗。

GPT-Image-2.5 Sunburst

GPT-Image-2.5 Sunburst 是 OpenAI 的图像生成与编辑模型，支持文本和图像输入、图像输出，适合对编辑精度要求较高的工作流。 支持 low、medium、high、xhigh、max 和 auto 六种质量设置。可通过 Image API 直接使用，也可作为 Responses API 图像生成工具的模型使用。 支持的图像接口： - 生成：`v1/images/generations` - 编辑：`v1/images/edits` 不支持音频和视频。 ### 官方价格 按每 100 万 Token 计费，单位为 USD： | 类型 | 输入 | 缓存输入 | 输出 | | --- | ---: | ---: | ---: | | 文本 | 5.00 | 1.25 | 不适用 | | 图像 | 8.00 | 2.00 | 30.00 | 模型仅输出图像，不收取文本输出费用。Token 单价与 GPT Image 2 相同，但 GPT Image 2 的计算器不能用于估算 GPT Image 2.5 的 Token 消耗。

Nano Banana Pro

Nano Banana Pro is a complex reasoning-driven engine for professional-grade image editing and generation, offering studio-level precision and advanced creative control. Nano Banana Pro is best suited for complex graphic design, high-fidelity product modeling, and factual data visualization that requires accurate text rendering and real-world grounding via Google Search.

Nano Banana Pro Preview

Nano Banana Pro 是 Google Gemini 的模型。 模型族：gemini-pro。 知识截止：2025-01。 模态：text、image → text、image。 上下文长度：65,536 tokens。

Nano Banana 2

Nano Banana 2 provides high-quality image generation and conversational editing at mainstream prices with low latency. It is the highly efficient counterpart to Gemini 3 Pro Image, optimized for speed and high-volume developer use cases. It supports 0.5K, 1K, 2K, and 4K resolutions, adds 1:4, 4:1, 1:8, and 8:1 aspect ratios, and supports image search grounding and chain of thought.

Nano Banana 2 Preview

Nano Banana 2 是 Google Gemini 的模型。 模型族：gemini-flash。 知识截止：2025-01。 模态：text、image、pdf → text、image。 上下文长度：65,536 tokens。

Nano Banana 2 Lite

Nano Banana Lite is the efficiency expert in the image generation family, providing ultra-low latency and highly cost-effective image generation and editing capabilities. Targeting an end-to-end latency of under 2 seconds, it significantly reduces TPU computing costs, making it ideal for high-concurrency interactive developer use cases and real-time consumer applications. It supports interleaved generation and editing from text to image-and-text, and from image to image-and-text, is optimized for 1K (1024x1024px) resolution, supports 14 aspect ratios, enables rapid multi-turn local editing, maintains high character consistency, and has SynthID + C2PA watermarks enabled by default.