Experiment with Gemini 2.0 Flash native picture technology

In December we first launched native picture output in Gemini 2.0 Flash to trusted testers. In the present day, we’re making it accessible for developer experimentation throughout all areas presently supported by Google AI Studio. You’ll be able to take a look at this new functionality utilizing an experimental model of Gemini 2.0 Flash (gemini-2.0-flash-exp) in Google AI Studio and by way of the Gemini API.

Gemini 2.0 Flash combines multimodal enter, enhanced reasoning, and pure language understanding to create photos.

Listed here are some examples of the place 2.0 Flash’s multimodal outputs shine:

1. Textual content and pictures collectively

Use Gemini 2.0 Flash to inform a narrative and it’ll illustrate it with photos, protecting the characters and settings constant all through. Give it suggestions and the mannequin will retell the story or change the model of its drawings.

Sorry, your browser would not help playback for this video

Story and illustration technology in Google AI Studio

2. Conversational picture modifying

Gemini 2.0 Flash helps you edit photos by way of many turns of a pure language dialogue, nice for iterating in direction of an ideal picture, or to discover completely different concepts collectively.

Sorry, your browser would not help playback for this video

Multi-turn dialog picture modifying sustaining context all through the dialog in Google AI Studio

3. World understanding

Not like many different picture technology fashions, Gemini 2.0 Flash leverages world data and enhanced reasoning to create the proper picture. This makes it good for creating detailed imagery that’s lifelike–like illustrating a recipe. Whereas it strives for accuracy, like all language fashions, its data is broad and common, not absolute or full.

Sorry, your browser would not help playback for this video

Interleaved textual content and picture output for a recipe in Google AI Studio

4. Textual content rendering

Most picture technology fashions battle to precisely render lengthy sequences of textual content, typically leading to poorly formatted or illegible characters, or misspellings. Inner benchmarks present that 2.0 Flash has stronger rendering in comparison with main aggressive fashions, and nice for creating commercials, social posts, and even invites.

Sorry, your browser would not help playback for this video

Picture outputs with lengthy textual content rendering in Google AI Studio

Begin making photos with Gemini at present

Get began with Gemini 2.0 Flash by way of the Gemini API. Learn extra about picture technology in our docs.

from google import genai
from google.genai import varieties

consumer = genai.Shopper(api_key="GEMINI_API_KEY")

response = consumer.fashions.generate_content(
    mannequin="gemini-2.0-flash-exp",
    contents=(
        "Generate a narrative a few cute child turtle in a 3d digital artwork model. "
        "For every scene, generate a picture."
    ),
    config=varieties.GenerateContentConfig(
        response_modalities=["Text", "Image"]
    ),
)

Whether or not you’re constructing AI brokers, creating apps with lovely visuals like illustrated interactive tales, or brainstorming visible concepts in dialog, Gemini 2.0 Flash permits you to add textual content and picture technology with only a single mannequin. We’re desperate to see what builders create with native picture output and your suggestions will assist us finalize a production-ready model quickly.