Python script: filter Sume video models by length, size and ratio

Yes, you can pick a Sume video model from code instead of memory. GET /v1/videos/models returns each row with its supported_* fields, so a filter on length, resolution and aspect ratio gives you the ids that can serve a request. The script below prints matching ids and exits non-zero when none match.

The script

It reads the key from SUME_API_KEY, accepts the three constraints as arguments and treats missing fields as no match.

import os, sys, requests

secs, res, ratio = int(sys.argv[1]), sys.argv[2], sys.argv[3]
r = requests.get(
    "https://api.sume.com/v1/videos/models",
    headers={"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"},
    timeout=30,
)
r.raise_for_status()
hits = []
for m in r.json().get("data") or []:
    if secs not in (m.get("supported_durations") or []):
        continue
    if res not in (m.get("supported_resolutions") or []):
        continue
    if ratio not in (m.get("supported_aspect_ratios") or []):
        continue
    hits.append(m["id"])
print("\n".join(hits) or "none")
sys.exit(0 if hits else 1)

What to do with the result

Run python pick.py 12 1080p 9:16 and you get the ids that list 12 seconds at 1080p in 9:16. Then check each row's other limits in the Video Router docs: references, audio and end frames differ per row. If the output is none, change one constraint at a time; length is usually the first to cut, since many rows stop at 10 or 15 seconds.

A caution on field names

The field names in the script follow the supported_* pattern the docs describe; print one row first with print(r.json()["data"][0]) to confirm the exact keys in your environment before you rely on the filter.