Replicate Wan 2.1 at $0.25/s vs Sume Wan 3.0: two different models
Replicate lists Wan 2.1 at $0.09/s (480p) and $0.25/s (720p). Sume bills Wan 3.0 at list x 1.25, $0.125/s at 720p; they are different model generations.
Replicate's pricing page lists Wan 2.1 at $0.09 per second for 480p and $0.25 per second for 720p. Sume serves Wan 3.0, a newer model, at the provider list price times 1.25, which is $0.125 per second at 720p, so the per-second numbers are not a like-for-like model comparison.
The rates
Sume's rate card lists Wan 3.0 at a provider price of $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p, marked subject to change. Multiply by 1.25 for what Sume bills; confirm the live numbers in GET /v1/catalog.
| Resolution | Replicate Wan 2.1 | Sume Wan 3.0 (list x 1.25) |
|---|---|---|
| 480p | $0.09 | $0.0625 |
| 720p | $0.25 | $0.125 |
| 1080p | not listed | $0.25 |
A 10-second 720p clip
On Replicate's listed rate that is $2.50 for Wan 2.1. On Sume it is $1.25 for Wan 3.0, which accepts 2 to 30 seconds per clip. Replicate's per-second figure is a time-based rate for that model, and the page describes billing by compute time, so actual cost follows how long a run takes.
Different model versions produce different video, so price is only one input; run the same prompt on both before you decide.
API shape
Sume uses POST /v1/videos with model: "wan-3.0", an OpenRouter-compatible body. seed and size are unsupported and provider.options is rejected. Results live on durable media.sume.com URLs, while Replicate's webhook documentation says outputs are deleted after an hour, so download promptly there.
Sources
Related posts
More in Comparisons
- Rev AI human transcription $1.99 a minute vs Sume STT $0.01 a minute
Rev AI lists human transcription at $1.99 a minute, 199 times Sume's machine STT at $0.01. What the price gap pays for and when machine output is enough.
- Rev AI Whisper (English) $0.005 a minute, foreign $0.30 an hour
Rev AI lists Whisper Large (English) at $0.005 a minute and Reverb Foreign Language at $0.30 an hour; Sume STT is $0.01 a minute. Per-hour table and rounding.
- Riverside Pro $24 with 15 hours of tracks vs Sume audio detach
Riverside Pro lists 15 hours of separate track downloads at $24 a month, billed annually. Sume audio detach pulls one track from a video for $0.01.
- Runware bills only successful requests: how Sume handles failures
Runware says you pay only for successful requests; its rates assume default settings. Sume reserves at submit and refunds failed jobs.
Written by Sume