Gpt-image-1.5 image creator and editor app - Python, desktop UI

Preliminary GPT-Image-2 support is merely by editing the truth table in code.

Find gpt-image-1.5 in MODEL_SPECS, and add the new model above the existing entries.

MODEL_SPECS: dict[str, dict[str, Any]] = {
    "gpt-image-2": {
        **GPT_IMAGE_MODEL_SPEC_BASELINE,
        "order": 8,
        "sizes": [
            "auto",
            "1024x1024",
            "1536x1024",
            "1024x1536",
            "1280x720",
            "1920x1088",
        ],
        "supports_input_fidelity": False,
        "pricing": {
            # Prices are per 1M tokens
            "text_input_price_per_million": 5.00,
            "text_output_price_per_million": 0.00,
            "image_input_price_per_million": 8.00,
            "image_output_price_per_million": 30.00,
        },
    },
    "gpt-image-1.5": {
        **GPT_IMAGE_MODEL_SPEC_BASELINE,
        "order": 10,
        "supports_input_fidelity": True,
        ...

Add your own resolutions at 16 pixel size increments.

It will take a bit more to make a good UI for arbitrary image resolution picking and validating minimum and maximum token counts. I’ll probably just copy the behavior of image loading dialog, where you can type your own canvas sizes into the drop-down, and then that text value needing the format "1234x1224" would be validated and tweaked. The rest of the code is forward-looking, anticipating other resolutions and aspect ratios.