Gemini 3.1 Flash Image — fast image generation

Use gemini-3.1-flash-image-preview for fast creative iteration and draft variants. Call it through the OpenAI-compatible Images endpoint (POST https://api.quotaflow.ai/openai/v1/images/generations) or natively through Gemini generateContent (POST https://api.quotaflow.ai/v1beta/models/gemini-3.1-flash-image-preview:generateContent). Both endpoints are served by the same image product at the same per-image price.

What is Gemini 3.1 Flash Image?

Gemini 3.1 Flash Image is the fast image model in the Quotaflow creative catalog. It is a good first pass when you need many variants or quick feedback before choosing a final candidate.

Why use Gemini 3.1 Flash Image?

Use this model when speed and iteration count matter. Product teams can generate several concepts, review them quickly, and reserve slower or higher-cost image paths for final candidates.

Curated models

Key features

• Fast drafts: ideal for initial concepts and many variants.
• OpenAI-compatible option: call /openai/v1/images/generations from OpenAI-style clients.
• Gemini-native option: call :generateContent from Gemini SDKs and Gemini-style clients.
• Image edits: send 1 to 14 reference images to /openai/v1/images/edits, as multipart image[] parts or JSON images[].image_url data URLs, or as inlineData parts on :generateContent.

How to generate images

OpenAI-compatible generation:

curl https://api.quotaflow.ai/openai/v1/images/generations \
  -H "Authorization: Bearer $QUOTAFLOW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-flash-image-preview",
    "prompt": "A crisp social media image of a teal AI dashboard floating above a city map.",
    "size": "1024x1024",
    "response_format": "b64_json"
  }'

Sizes are auto, 1024x1024, 1024x1536 and 1536x1024. Gemini-specific fields such as aspect_ratio or image_config are rejected on this endpoint with HTTP 400 naming the field in param; the Images page documents the full request contract.

Gemini-native generation:

curl "https://api.quotaflow.ai/v1beta/models/gemini-3.1-flash-image-preview:generateContent" \
  -H "x-goog-api-key: $QUOTAFLOW_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          { "text": "A crisp social media image of a teal AI dashboard floating above a city map." }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": { "aspectRatio": "1:1", "imageSize": "1K" }
    }
  }'

The native request accepts one user turn of text parts and inlineData reference-image parts, plus generationConfig.responseModalities (must include IMAGE), generationConfig.imageConfig and generationConfig.candidateCount (1). Only aspectRatio 1:1 with imageSize 1K (1024x1024) is available until more sizes are measured; other ratios and sizes, and every other field, return HTTP 400 INVALID_ARGUMENT with a message naming the field. :streamGenerateContent is not available for this model. The image comes back in Google's native shape, in candidates[0].content.parts[0].inlineData. The Images page documents the full native contract.

Usage accounting

For Gemini image models, usage reports Quotaflow's own metered output token count (output_tokens and total_tokens), the same source used for billing. It does not include input_tokens because this product is priced per image, and it never uses the vendor's self-reported token count. On the native endpoint the same count is reported in usageMetadata as candidatesTokenCount, totalTokenCount and candidatesTokensDetails; promptTokenCount is omitted for the same reason.

Pricing

This model is a good starting point when you expect repeated creative iteration. Confirm key access and package limits before exposing high-volume generation to end users.

Use cases

FAQ

Should I choose Flash Image or Pro Image?

Start with Flash Image for speed and iteration. Use Pro Image when final quality or complex instruction following matters more than latency.

Can I use the OpenAI-compatible endpoint?

Yes. Use /openai/v1/images/generations for generation and /openai/v1/images/edits to edit with reference images.

Can I use a Gemini SDK or generateContent?

Yes. Call :generateContent with the Gemini request shape; send reference images as inlineData parts. Streaming is not available for image generation.

Next steps

• Use GPT Image 2 for OpenAI-compatible image-first apps.
• Browse the Model Library.