Gemini 3 Pro Image API (formerly Preview): Pricing, Resolutions, and How to Call It

The Gemini 3 Pro Image API (formerly Preview) — widely recognized across developer communities as Nano Banana Pro — is the premium image generation and editing engine in Google's Gemini family. As teams transition AI visual features from internal demos to commercial production, it frequently becomes the default standard.

This guide covers Google's official GA naming evolution, Pixapi routing aliases, exact resolution tiers, aspect ratio syntax, and pricing math.

What is Gemini 3 Pro Image?

Gemini 3 Pro Image is Google's flagship visual model for high-fidelity text-to-image synthesis and composition-preserving image editing. It natively supports up to 4096×4096 (4K) output and excels at photorealistic edits where existing subject positions and lighting must remain intact.

Key details regarding naming and integration:

1. Google GA transition: While early beta releases used gemini-3-pro-image-preview, Google has formalized the model under its GA production name gemini-3-pro-image.
2. Pixapi compatibility routing: To ensure seamless operation, Pixapi recommends the official GA ID gemini-3-pro-image while actively maintaining the classic gemini-3-pro-image-preview alias.
3. Nano Banana Pro: Nano Banana Pro is a community nickname commonly associated with Gemini 3 Pro Image.

Gemini 3 Pro Image API pricing

Pixapi bills transparently in credits where 1 credit = $0.01, matching USD costs 1:1:

Prices verified August 2026. Because 2K output is billed at the same price as 1K ($0.08), it serves as the sweet spot for production web applications. At 4K, Pixapi charges $0.16 compared to Google's standard rate of ~$0.24, delivering roughly 33.3% in infrastructure savings. Current credit packages are available on the pricing page.

For broader budgeting context across the Gemini image family:

How to call the API

Authentication uses a Bearer API key generated in your Pixapi dashboard. Important: pass desired aspect ratio (1:1, 16:9, 9:16, 4:3) in size, and select resolution tiers (1K, 2K, 4K) via the model ID.

1. Text-to-image example: 2K square render

curl -X POST https://api.pixapi.ai/v1/images/generations \
  -H "Authorization: Bearer $PIXAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-pro-image-2k",
    "prompt": "Studio product photo of a ceramic coffee mug on linen, soft daylight",
    "size": "1:1"
  }'

2. Image edits example

For image-to-image and secondary editing, send reference URLs to the dedicated edits endpoint:

curl -X POST https://api.pixapi.ai/v1/images/edits \
  -H "Authorization: Bearer $PIXAPI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3-pro-image",
    "image": "https://example.com/input.jpg",
    "prompt": "Replace the background with a marble countertop, keep the mug unchanged"
  }'

Full parameter documentation and schema details live in the image generation API reference. Always keep your API key secure server-side.

When to pick Pro over Flash

When planning architecture around Gemini visual APIs:

• Pro ($0.08) — final deliverables: primary marketing banners, commercial product photos, and print-ready high-detail imagery.
• Flash ($0.04) — iterative prototyping: draft exploration, variant generation, and real-time interactive user flows.
• Flash Lite / 2.5 Flash ($0.02) — bulk background jobs: avatars, batch icon transformations, and thumbnail generation.

A popular, cost-efficient strategy is presenting candidates to users via Flash, then running the confirmed favorite through Pro for 2K or 4K final delivery.

Practical tips for production deployment

1. Explicit resolution targeting: Pixapi consistently routes default IDs (like gemini-3-pro-image) to 1K, -2k to 2K, and -4k to 4K.
2. Standardize on 2K: Because 2K is priced identically to 1K ($0.08), default your rendering requests to 2K for noticeably sharper output at zero extra cost.
3. Use 4K intentionally: 4K costs 16 credits ($0.16) and is best reserved for hero banners or zoomable high-resolution inspection.

Related reading

• GPT Image 2 vs Nano Banana: which should you use?
• Seedance API pricing explained
• Model catalog and Nano Banana