Nova Reel 2-minute videos in 6 s steps vs Sume's 30 s video jobs
Nova Reel makes up to two minutes through async invoke in 6-second increments. Sume's catalog tops out at 30 seconds per job, so a long video needs several.
Nova Reel can return one video up to two minutes long; no Sume video model in the docs does. AWS's user guide says Nova Reel generates in 6-second increments up to 2 minutes and runs only through the asynchronous StartAsyncInvoke call. Sume's video docs list 30 seconds as the longest single clip, for seedance-2.5 (4 to 30) and wan-3.0 (2 to 30).
Side by side
Both services are asynchronous, so the shape of your code survives a move: submit, keep an id, poll or take a callback, then fetch. What changes is the length of one job and where the output lands.
| Item | Amazon Nova Reel | Sume /v1/videos |
|---|---|---|
| Longest single output | 2 minutes | 30 seconds (seedance-2.5, wan-3.0) |
| Duration steps | 6-second increments | whole seconds from the model's supported_durations |
| Invocation | StartAsyncInvoke only | POST /v1/videos, 202 with an id and polling_url |
| Model id | amazon.nova-reel-v1:1 | catalog ids such as seedance-2.5 |
| Learn the result | Async invoke, then read the output when done | Poll, or send callback_url for a signed webhook |
Planning a long video
A 2-minute target is at least four 30-second jobs on Sume, and the clips will not match each other unless you plan for it. Sume does not join generated clips in /v1/videos; assembly is a separate step, and the Timeline 1.0 docs describe the Sume surface that sequences clips. Check them for limits before you commit to a design.
Cost scales with the sum of the clips, and the submit reserves provider list price times 1.25 for each. Read pricing_skus from GET /v1/videos/models and multiply before you queue a long batch.
What I did not compare
I have no quality or latency numbers for either side, and none are implied here. Pick on length, price and the controls you need, then test with a real prompt.
Sources
Related posts
More in Comparisons
- One Sume image upscale costs the same as 22 seconds of video upscale
Sume image upscale is $0.20 per image and video upscale is $0.009 per second, so one image upscale buys 22.2 seconds of video. A cutout is $0.0225.
- OpenAI Image API vs Sume Image API for GPT Image 2.5: what differs
Same model family, different request surface. Moderation, revised_prompt, streaming, output_compression and masks compared between OpenAI direct and Sume.
- OpenRouter video client on Sume: six documented differences
Sume says /v1/videos matches the OpenRouter Video Generation API except six items: base path, auth, model ids, auto routing, size and provider.options.
- Pick an image model by job: product photo, poster text, edit
Five new image releases in a month. A job-first table for Sume: GPT Image 2.5, Ideogram 4.5, Nano Banana 2.1, FLUX.2 and Qwen, with the limits that decide.
Written by Sume