← All models

Gemini 3.1 Flash Image

Published Publisher Googlegemini-3.1-flash-image

Overview

Gemini 3.1 Flash Image, also called Nano Banana 2, is a multimodal image-generation and conversational-editing model optimized for low latency and high-volume use. It is positioned as a more efficient counterpart to Gemini 3 Pro Image. (ai.google.dev)

Release

Google introduced Nano Banana 2 on February 26, 2026. The stable model gemini-3.1-flash-image became generally available on May 28, 2026, including through Gemini Enterprise Agent Platform. The supplied date refers to the stable release, not the model’s first announcement. (blog.google)

Capabilities and uses

  • Accepts text, images, video and PDF input through the Gemini API, and produces images and text. It supports reasoning and grounding with Google Search. (ai.google.dev)
  • Generates and edits images conversationally; supports 0.5K, 1K, 2K and 4K output, with 1K as the default. It can use search results to inform generation and offers improved text rendering and aspect-ratio control. (ai.google.dev)
  • Suitable for visual content production, marketing mockups, infographics, diagrams, product imagery and video-derived thumbnails. (blog.google)

Limits and availability

  • Maximum input: 131,072 tokens; maximum output: 32,768 tokens. (ai.google.dev)
  • Gemini API documentation lists audio generation, structured output, function calling, code execution and caching as unsupported. Platform capabilities differ: Agent Platform separately lists implicit context caching as supported. (ai.google.dev)
  • The requested number of images is not always followed exactly. Search grounding does not currently support using real-world images of people from web search; generated images include a SynthID watermark. (ai.google.dev)
  • At the May 28 Agent Platform release, video input and 4K output were described as preview features; check current platform documentation for their status. (docs.cloud.google.com)

Pricing

Gemini Developer API paid tier, queried November 16, 2026; prices are in USD per 1 million tokens unless otherwise stated. These are Gemini Developer API prices, not necessarily Google Cloud Agent Platform prices. (ai.google.dev)

Mode Input: text and image Output: text and thinking Output: image
Standard $0.50 $3.00 $60.00
Batch $0.25 $1.50 $30.00

Standard image-output equivalents are approximately $0.045 per 0.5K image, $0.067 per 1K image, $0.101 per 2K image and $0.151 per 4K image. Batch equivalents are approximately $0.022, $0.034, $0.050 and $0.076 per image, respectively. Input and other output tokens are billed separately. Video input pricing was not verified on the cited Gemini Developer API pricing table. (ai.google.dev)

For Google Web and Image Search grounding, the paid tier includes 5,000 free search requests per month shared across Gemini 3 and newer models, then charges $14 per 1,000 search requests. A submitted model request may trigger multiple search queries. The listed free API tier is unavailable for this model, though testing in Google AI Studio is available. (ai.google.dev)

Official sources