Image launches this week vs Sume's catalog: FLUX 3, Ideogram 4.5

Four image launches landed in the last two weeks on Higgsfield's changelog: GPT Image 2.5 on Sep 22, Seedream 5.0 Flash on Sep 23, Ideogram 4.5 on Sep 30 and FLUX 3 Image on Oct 1. Sume's image catalog carries GPT Image 2.5 (both variants), Ideogram 4.5 and Seedream 5 Lite. It does not carry FLUX 3 Image or Seedream 5.0 Flash today. Dates are Higgsfield's; other vendors may date launches differently, and GET /v1/images/models is the live source for the Sume column.

The table

Dates are from Higgsfield's changelog (a distributor), read 2026-10-05. Prices are from each vendor page we could read, or the Sume docs where noted.

What each launch is for

Short notes from the Higgsfield entries, read 2026-10-05, not tested by us:

• GPT Image 2.5: Flare and Sunburst variants, per Higgsfield and OpenAI's guide (Sunburst for precise edits, Flare for fast everyday generation).
• Seedream 5.0 Flash: fast tier with multi-reference style blending, generate and edit.
• Ideogram 4.5: repeated edits that keep colours, textures and details, reference edits, text replacement, transparent backgrounds, 1K or 2K.
• FLUX 3 Image: text-to-image and image-to-image, multilingual text, 768p to 4K, native 2K and 4K.

What to do with the gaps

For FLUX 3 Image or Seedream 5.0 Flash, call the vendor or a distributor directly today; the Sume catalog may add rows later, so check the models endpoint instead of assuming. For the two that are in the catalog, the request shape is the same one POST /v1/images, which means you can switch on the model field alone. Check the supported parameters for each row, because an unsupported one returns 400 unsupported_parameter.

import asyncio, os, httpx

async def main():
    headers = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}
    async with httpx.AsyncClient(timeout=30) as c:
        r = await c.get("https://api.sume.com/v1/images/models", headers=headers)
    r.raise_for_status()
    ids = {m["id"] for m in r.json()["data"]}
    for want in ("openai/gpt-image-2.5", "ideogram/ideogram-v4.5", "bytedance-seed/seedream-5-lite"):
        print(want, want in ids)
    print([i for i in ids if "flux" in i])

asyncio.run(main())

Pick by job

Text on an ad: Ideogram 4.5 or GPT Image 2.5 (test both on the same copy). Masked local edit: ChatGPT Image 2.5, the only Sume row with a mask_url. Big prints: see the cost of a 16 MP upscale on Sume. For an ad-text comparison with FLUX 3, see FLUX 3 or Ideogram 4.5 for ad text.

Caveats

Higgsfield is a distributor, so its dates are when it added each model, not necessarily when the vendor launched it. One press report dated FLUX 3 Image to October 2; BFL's own pages we read give no launch date. Treat the dates as approximate and the Sume column as live data to re-check. Vendors change prices, so re-read each pricing page before you commit to a plan.