Wan 3.0 clip lengths that fit a $1.00 scheduled run cap
A schedule without its own cap gets $1.00 per run. At Wan 3.0 rates that buys 16 s at 480p, 8 s at 720p or 4 s at 1080p; here is the table.

A Sume schedule with no cap of its own runs under a $1.00 default generation cap, and at Wan 3.0 list prices that default buys exactly 16 seconds at 480p, 8 seconds at 720p, or 4 seconds at 1080p. A 10-second 720p clip costs $1.25, so it is over the default before the run starts.
What $1.00 buys on Wan 3.0
Wan 3.0 accepts 2 to 30 seconds per clip. The per-second rate doubles with each resolution step, so the longest clip that fits the default halves each time. Totals below are rate times seconds, from the Sume price tool.
| Resolution | Rate per second | Longest clip under $1.00 | Cost of that clip | Cost of a 10 s clip |
|---|---|---|---|---|
| 480p | $0.0625 | 16 s | $1.00 | $0.625 |
| 720p | $0.125 | 8 s | $1.00 | $1.25 |
| 1080p | $0.25 | 4 s | $1.00 | $2.50 |
Lowering the cap per run
The Scheduled docs say the default cap is $1.00 when none is set, and that a caller may pass generation_spend_cap_usd on an API-triggered run. The API clamps a number to min(request, schedule cap), so it can only lower the ceiling. To keep a 480p test run at about half the default, send 0.5:
curl -sS -X POST "https://api.sume.com/v1/actions/$ACTION_ID/runs" \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: $(uuidgen)" \
-d '{"generation_spend_cap_usd": 0.5}'Limits and gotchas
Do not plan around raising the cap per run. A number above the schedule cap is clamped down to it, and 0 is rejected with a 400. Sending null removes the automation ceiling, so wallet balance and admission limits are then the only brake; that is rarely what an unattended schedule should do.
- The Developer API can list, read and run schedules, but not create or edit them, so the schedule cap itself is set where you author the schedule.
- A 2-second Wan 3.0 clip at 1080p costs $0.50, so two of them reach the default exactly.
- Check the model's own limits with the model list before you pick seconds; they differ per model.
Sources
Related posts
More in Agents
- Your agent calls generate_video, or Sume's agent does: two meters
With hosted MCP your model client runs the loop and you pay its provider; with an Agent Completion Sume's agent runs it and you set generation_spend_cap_usd.
- Planlock in front of Sume MCP: approve the plan, then the calls
Planlock is an MCP proxy that enforces a human-approved plan. Where it fits in front of Sume's hosted MCP, and which Sume gates still do work behind it.
- Run the Sume video agent from your backend with Agent Completions
POST /v1/agent/completions runs the same agent as the Sume Agents chat, with tools and media generation, and returns an async run receipt you poll or webhook.
- Safe automation for AI agents that call paid APIs
Keep agents read-only by default, keep secrets out of logs, and on hosted MCP send an idempotency_key, preview with dry_run, and cap with max_spend_usd.
Written by Sume