Luma Ray 3.2 20 seconds 1080p: how Sume's catalog limits read

Luma says Ray3.2 makes up to 20 seconds at 1080p. In Sume's video docs, seedance-2.5 is the one model stated to accept 20 seconds at 1080p.

4 min readSume
All posts

Luma's Ray3.2 announcement, dated June 9, 2026, lists 1080p and up to 20 seconds per generation. Sume lists no Luma model, but its video docs state that seedance-2.5 accepts 4 to 30 seconds at 480p, 720p and 1080p, so a 20-second 1080p request fits that model's stated range. No other model in the docs read is stated to cover both.

What does Luma state about length and resolution?

Luma's announcement, read 2026-09-29, gives 1080p and up to 20 seconds per generation. It gives no price or minimum length in the text saved for this page.

Which Sume models take 20 seconds at 1080p?

Limits are per model, and the docs say so directly: they are not uniform. Read GET /v1/videos/models for the live values before you commit to a plan.

Duration and resolution limits stated in Sume's video docs, read 2026-09-29.
ModelDurationResolutions stated
seedance-2.54–30 seconds480p, 720p, 1080p
seedance-24–15 seconds1080p offered
wan-3.02–30 secondsNot stated in the passage read
minimax-h35–15 secondsNative 480p and 768p
minimax-h3-max5–15 seconds480p, 768p, 1080p (a latent refinement from native 768p)
gemini-omni-flash-1.13–10 seconds360p, 720p, 1080p, 4K

What does that mean for a 20-second 1080p request?

Only seedance-2.5 is stated to accept both 20 seconds and 1080p. wan-3.0 accepts 20 seconds, but the docs passage read does not state its resolutions. minimax-h3-max reaches 1080p but stops at 15 seconds, and the rest of the catalog tops out at 15 seconds.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: clip-20s-1080p" \
  -d '{
    "model": "seedance-2.5",
    "prompt": "A slow dolly along a rain-lit street at night",
    "resolution": "1080p",
    "duration": 20,
    "aspect_ratio": "16:9",
    "mode": "async"
  }'

What if I need more than the limit?

Chain clips. The longest single request in the catalog is 30 seconds, so anything longer is several clips joined with Timeline 1.0; the steps are in make a 1-minute AI video. For details on the request fields, see the video docs.

How many clips does a longer video take?

Divide the length you want by the clip limit and round up. That is arithmetic, not a Sume rule, and it shows why a 30-second ceiling matters for anything past half a minute. The clips are then joined with Timeline 1.0.

Clips needed for a 60-second video by per-clip limit, arithmetic on limits stated in the docs, 2026-09-29.
Per-clip limitExample modelClips for 60 seconds
30 secondsseedance-2.52
20 secondsA model with a 20-second limit3
15 secondsminimax-h34
10 secondsgemini-omni-flash-1.16

How do I read the limits from code?

The catalog response carries supported_durations and supported_resolutions for each model. Filter it for a duration of 20 and a resolution of 1080p rather than trusting a blog table, including this one; the docs page is prose and a model can change.

Sources

Related posts

More in Models

All Models posts

Written by Sume