fal lists Seedream 5.0 and Krea 2: find what Sume's image catalog has
fal.ai/models names GPT Image 2.5, Flux 2, Nano Banana 2, Ideogram 4, Krea 2 and Seedream 5.0. A Python check of which names appear in Sume's /v1/images/models.

fal.ai/models lists Seedream 5.0, GPT Image 2.5, Flux 2, Nano Banana 2, Ideogram 4 and Krea 2 among its image models. Which of those you can call through Sume is answered by GET /v1/images/models, not by this page, and the script below prints the matches for each name.
Sume's docs confirm two of the six by name: GPT Image 2.5 and an Ideogram release, the latter as 4.5.
What each source confirms
Names from the fal listing, and what the Sume Image API page says about them, read 2026-10-03.
| Name on fal's list | Named on Sume's Image API page | Sume id given there |
|---|---|---|
| GPT Image 2.5 | Yes, Flare and Sunburst | openai/gpt-image-2.5, openai/gpt-image-2.5-sunburst |
| Ideogram 4 | Ideogram 4.5 is named | ideogram/ideogram-v4.5 |
| Seedream 5.0 | Page example uses Seedream 4.5 | bytedance-seed/seedream-4.5 in examples |
| Flux 2 | Example uses a FLUX.2 id | black-forest-labs/flux.2-pro in an example |
| Nano Banana 2 | Not checked on this page | Read the catalog |
| Krea 2 | Not checked on this page | Read the catalog |
Check the live catalog
The script fetches the image catalog and searches model ids and names for each keyword. A keyword with no hit means the catalog does not list it today, not that it never will. Names are matched case-insensitively and loosely because ids use provider slugs.
import os, requests
WANTED = ["seedream", "gpt-image-2.5", "flux", "banana", "ideogram", "krea"]
r = requests.get("https://api.sume.com/v1/images/models",
headers={"Authorization": "Bearer " + os.environ["SUME_API_KEY"]},
timeout=60)
r.raise_for_status()
models = r.json()["data"]
for word in WANTED:
hits = [m["id"] for m in models
if word in m["id"].lower() or word in (m.get("name") or "").lower()]
print(f"{word:<14}", hits or "no match in catalog")How to use the result
Treat a model name as a search term until the catalog turns it into an id.
- Once you have an id, read its capabilities from
/v1/images/models/{model_id}/endpointsbefore pinning size, quality or references. - A parameter the model does not list returns
400 unsupported_parameterinstead of being dropped, so a wrong assumption fails fast. - If a name is missing, use
sume/autoto let Image Router choose, or pick the closest listed id. - Re-run the check when a vendor announces a new version, since catalogs change independently of listings.
A note on version names
Version numbers drift between lists: a listing may say Ideogram 4 while the product page says 4.5, and Sume's page names 4.5. Compare ids, not marketing names, and keep the id in your config rather than a label in your prompt.
Sources
Related posts
More in Integrations
- LangGraph interrupt or wait: resume a graph from a Sume webhook
A LangGraph interrupt is for a human decision, not a video render. Pause for approval, then resume the graph from a signed Sume webhook keyed by job_id.
- Mastra eager tool execution: dry-run Sume calls first
Mastra 1.71 can start a tool once its own arguments are complete. For paid Sume generation that means a stable idempotency_key and a dry run before any spend.
- Mastra 1.72 crash recovery and leases: checkpoint the Sume job id
Mastra 1.72 adds multi-worker task leases and crash recovery. Store the Sume job id before waiting so a recovered worker polls instead of paying twice.
- n8n 2.41.6 task runner and a Sume webhook verifier that never throws
n8n 2.41.6 keeps its task runner alive on unhandled rejections. Write the Sume webhook check so a bad signature returns false; answer 2xx only after storing.
Written by Sume