Générer des images avec Gemini

POST

v1beta

models

{model}

:generateContent

import os
from google import genai
from google.genai import types

client = genai.Client(
    http_options={"api_version": "v1beta", "base_url": "https://api.cometapi.com"},
    api_key=os.environ.get("COMETAPI_KEY"),
)

response = client.models.generate_content(
    model="gemini-3.1-flash-lite-image",
    contents="A Monarch butterfly anatomical sketch on textured parchment",
    config=types.GenerateContentConfig(
        response_modalities=["TEXT", "IMAGE"],
        image_config=types.ImageConfig(aspect_ratio="1:1", image_size="1K"),
    ),
)

final_image = None
for part in response.parts:
    if getattr(part, "thought", False):
        continue
    if part.text:
        print(part.text)
    elif image := part.as_image():
        final_image = image

if final_image:
    final_image.save("output.png")

{
  "candidates": [
    {
      "content": {
        "role": "<string>",
        "parts": [
          {
            "text": "<string>",
            "inlineData": {
              "mimeType": "<string>",
              "data": "<string>"
            },
            "thought": true,
            "thoughtSignature": "<string>"
          }
        ]
      },
      "index": 123,
      "safetyRatings": [
        {
          "category": "<string>",
          "probability": "<string>"
        }
      ]
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 123,
    "candidatesTokenCount": 123,
    "totalTokenCount": 123,
    "thoughtsTokenCount": 123
  }
}

Pour des exemples étape par étape, consultez Use Gemini image models.

Les options de génération d’images Gemini peuvent évoluer à mesure que Google met à jour les modèles d’image et generateContent. Consultez la documentation sur la génération d’images Gemini pour obtenir la liste complète la plus récente des paramètres et le comportement spécifique au fournisseur.

gemini-2.5-flash-image doit être arrêté par le fournisseur le 2026-10-02. Pour les nouveaux projets, utilisez gemini-3.1-flash-image-preview ou gemini-3-pro-image-preview. Consultez le calendrier de dépréciation de Google pour plus de détails.

Les réponses d’image Gemini peuvent inclure des parties d’image intermédiaires où thought vaut true. Il ne s’agit pas de la sortie finale. Lors de l’enregistrement des images générées, ignorez les parties thought: true et utilisez la dernière partie d’image où inlineData existe et thought n’est pas true.

Autorisations

Authorization

string

header

requis

Use your CometAPI API key.

Paramètres de chemin

model

string

défaut:gemini-3.1-flash-image-preview

requis

The Gemini image model to use. See the Models page for current options.

Corps

application/json

contents

object[]

requis

Conversation turns. Each item has a role ("user" or "model") and parts array containing text and/or image data.

Show child attributes

generationConfig

object

Controls generation behavior — output modalities, image resolution, thinking, etc.

Show child attributes

tools

object[]

Optional tools. Pass [{"google_search": {}}] to enable Google Search grounding for real-time information in generated images.

Show child attributes

Réponse

200 - application/json

Success

candidates

object[]

Show child attributes

usageMetadata

object

Show child attributes

Create image edit

Utiliser les modèles d’image Gemini

import os
from google import genai
from google.genai import types

client = genai.Client(
    http_options={"api_version": "v1beta", "base_url": "https://api.cometapi.com"},
    api_key=os.environ.get("COMETAPI_KEY"),
)

response = client.models.generate_content(
    model="gemini-3.1-flash-lite-image",
    contents="A Monarch butterfly anatomical sketch on textured parchment",
    config=types.GenerateContentConfig(
        response_modalities=["TEXT", "IMAGE"],
        image_config=types.ImageConfig(aspect_ratio="1:1", image_size="1K"),
    ),
)

final_image = None
for part in response.parts:
    if getattr(part, "thought", False):
        continue
    if part.text:
        print(part.text)
    elif image := part.as_image():
        final_image = image

if final_image:
    final_image.save("output.png")

{
  "candidates": [
    {
      "content": {
        "role": "<string>",
        "parts": [
          {
            "text": "<string>",
            "inlineData": {
              "mimeType": "<string>",
              "data": "<string>"
            },
            "thought": true,
            "thoughtSignature": "<string>"
          }
        ]
      },
      "index": 123,
      "safetyRatings": [
        {
          "category": "<string>",
          "probability": "<string>"
        }
      ]
    }
  ],
  "usageMetadata": {
    "promptTokenCount": 123,
    "candidatesTokenCount": 123,
    "totalTokenCount": 123,
    "thoughtsTokenCount": 123
  }
}

Modération de contenu

Clés API

Autorisations

Paramètres de chemin

Corps

Réponse