Retired AI model id to Sume id: GPT Image, Gemini and Sora lookup

Keep a single lookup from retired vendor ids to Sume ids, with the shutdown date next to each. As of 2026-10-06: OpenAI ends gpt-image-1 on 2026-10-23 and gpt-image-1-mini, gpt-image-1.5 and chatgpt-image-latest on 2026-12-01. Google ended gemini-2.5-flash-image on 2026-10-02. OpenAI ended Sora 2 and the Videos API on 2026-09-24.

Only the OpenAI image rows have a vendor-named replacement that exists on Sume. The Gemini and Sora rows are choices, so they are marked as such.

The lookup table

The lookup in code

explain returns a message, so it can sit in an exception handler or a startup check.

RETIRED = {
    "gpt-image-1": ("2026-10-23", "openai/gpt-image-2.5-sunburst"),
    "gpt-image-1-mini": ("2026-12-01", "openai/gpt-image-2.5-sunburst"),
    "gpt-image-1.5": ("2026-12-01", "openai/gpt-image-2.5-sunburst"),
    "chatgpt-image-latest": ("2026-12-01", "openai/gpt-image-2.5-sunburst"),
    "gemini-2.5-flash-image": ("2026-10-02", "google/nano-banana-2"),
    "sora-2": ("2026-09-24", "seedance-2.5"),
    "sora-2-pro": ("2026-09-24", "seedance-2.5"),
}

def explain(model_id):
    hit = RETIRED.get(model_id)
    if not hit:
        return f"{model_id}: not in the retired list"
    date, sume_id = hit
    return f"{model_id} retired {date}; try {sume_id} on Sume"

if __name__ == "__main__":
    for m in ("gpt-image-1", "sora-2", "nano-banana-2"):
        print(explain(m))

Pin, then re-read the vendor pages

The video row names seedance-2.5 and kling-3 only as starting points; the video ids and limits are in the Sora port guide. For image ids the longer notes are in the gpt-image-1 shutdown post and the Gemini post. Dates can move, so refresh the table when you re-read the vendor pages.