Nano Banana Pro (Gemini 3 Pro Image Preview)

google/gemini-3-pro-image-preview
VisionJSON
by Google · 2025-11-20

Nano Banana Pro is Google’s most advanced image-generation and editing model, built on Gemini 3 Pro. It extends the original Nano Banana with significantly improved multimodal reasoning, real-world grounding, and...

ctx65.5K tokens
Inputimage + text
Outputimage + text
p50 TTFT4.29 s
INPUT/ 1M tokens
OUTPUT/ 1M tokens
p50 TTFT4.29 s7d
p95 TTFT4.29 s7d
TRAFFIC267tokens / 7d

This model is a preview release from Google of an upcoming Gemini 3 Pro variant focused on image understanding. It accepts image and text inputs and generates text outputs. The "Nano Banana Pro"…

What is Nano Banana Pro (Gemini 3 Pro Image Preview)?

Who should use this preview model?

What are the input and output modalities?

How does OrcaRouter provide access to this model?

Code samples

Call from any SDK

  • Gemini SDKhttps://api.orcarouter.ai

Supported parameters

  • max_tokens
  • response_format
  • seed
  • stop
  • structured_outputs
  • temperature
  • top_p

Pricing

Per request$0.2400
CurrencyUSD
Flat fee per API call (image generation models)

Performance

p50 TTFT
4.29 s
Output speed
Collecting…
p95 TTFT
4.29 s
Error rate
0%

Public benchmarks

Ranking Distribution89777 tournaments
34718
First
26659
Second
16696
Third
11704
Fourth
Category Performance
Graphic DesignElo: 1283ImageElo: 1271ImageeditingElo: 1271LogoElo: 1259
Graphic DesignTop 22%
1283Elo
66.2%Win
29.0sAvg
Top 22%Rank
ImageTop 33%
1271Elo
62.3%Win
25.1sAvg
Top 33%Rank
ImageeditingTop 33%
1271Elo
66.3%Win
21.3sAvg
Top 33%Rank
LogoTop 33%
1259Elo
61.1%Win
24.9sAvg
Top 33%Rank
Source: Design Arena

FAQ

How much does it cost to use Nano Banana Pro through OrcaRouter?
Pricing is set by OrcaRouter per token for the google provider. There are no publicly listed per‑model rates for this preview. Visit OrcaRouter's pricing page or contact support for current input and output token costs. Image tokens count more than text tokens.
What is the context window size?
The model supports up to 65,536 tokens total (input + output). This includes tokens from both images and text. High‑resolution images quickly consume the context, so plan accordingly.
What are the main strengths of this preview model?
Strengths include accurate image description, visual question answering, and document understanding. It handles multiple images per request and integrates image and text reasoning in a single pipeline. It is a preview of Google's next‑generation multimodal architecture.
How does it compare to GPT‑4V or Claude 3 Vision?
Direct comparisons are limited because this is a preview. It has a smaller context window (65k vs 128k for GPT‑4V) and no published benchmarks. It may offer different trade‑offs in reasoning style and cost. Run your own tests for your specific use case.
How are images handled in terms of data privacy?
Image data is sent to OrcaRouter and then to Google's API. OrcaRouter's privacy policy applies. Images are processed in memory and not stored by Google unless explicitly logged. Do not send sensitive or personally identifiable information unless you have appropriate safeguards.
Can I call this model using the OpenAI Python library?
Yes. Change the base_url to https://api.orcarouter.ai/v1 and api_key to your OrcaRouter key. Set model to "google/gemini-3-pro-image-preview". The message format (text and image_url parts) is identical to OpenAI's vision API.
Is this model suitable for production use?
As a preview, it is not recommended for production until the final Gemini 3 Pro release. You may encounter bugs, higher latency, or sudden availability changes. Use it for prototyping and evaluation only.
Does the model support JSON mode or structured outputs?
The API supports the standard OpenAI response_format parameter (json_object). However, the model's ability to reliably output valid JSON may be inconsistent due to its preview status. Test thoroughly.
What image formats are supported?
Common formats like JPEG, PNG, WebP, and GIF (static frames only) are supported. The model does not handle video or animated GIFs. Maximum image resolution is not specified; very large images will be downscaled by Google's preprocessing.
Can I fine‑tune this model?
No. Fine‑tuning is not available for this preview model. Only the base model accessible via API is offered. For custom tuning, consider production models that support fine‑tuning on OrcaRouter.

Embed this badge

Google: Nano Banana Pro (Gemini 3 Pro Image Preview)pricing pending4290ms p50via OrcaRouter
HTML <a href="https://www.orcarouter.ai/models/google/gemini-3-pro-image-preview" target="_blank"> <img src="https://www.orcarouter.ai/embed/google/gemini-3-pro-image-preview.svg" alt="Google: Nano Banana Pro (Gemini 3 Pro Image Preview) on OrcaRouter" /> </a>
Markdown [![Google: Nano Banana Pro (Gemini 3 Pro Image Preview)](https://www.orcarouter.ai/embed/google/gemini-3-pro-image-preview.svg)](https://www.orcarouter.ai/models/google/gemini-3-pro-image-preview)

Model card as data

GET /api/public/models/google/gemini-3-pro-image-previewOpen
Machine-readable:/llms.txt/llms-full.txt