Free AI video API calls on Sume: unbilled checks, flat fees
Which Sume API calls are unbilled, which routes need no API key, and which paid calls charge one flat amount per call instead of a metered rate.
On the Sume API, the Timeline plan, the video filter check, video frames, and video inspect without a transcript are unbilled, requests that never run cost nothing, and six public routes need no API key. Generation itself is paid: each accepted submit reserves an estimate from your balance, and a few paid tools charge one flat amount per call instead of a metered rate.
The list below comes from Sume's API reference, Timeline 1.0, Video filter, Video frames, Video inspect, and Errors and spend docs, with prices read from the code behind API pricing on 2026-09-27. To preview a metered cost, see Estimate AI video generation cost; plans and the wallet are in How Sume pricing works.
Which Sume API calls cost nothing?
The docs mark the first four calls as unbilled or free. The hosted MCP preview checks admission and cost without submitting the job.
| Call | What the docs say | Explained in |
|---|---|---|
POST /v1/timeline-1.0/plan | Unbilled: no job, no credits reserved, no media downloaded. | Validate a timeline |
POST /v1/video-filter/check | Unbilled: no job, no credits reserved, no encoder. | ffmpeg filter allowlist |
POST /v1/video-frames | Unbilled: no seat and no reservation. | Extract video frames |
POST /v1/video-inspect without transcribe | Probe and stills are unbilled; only a transcript reserves. | Video inspect |
Hosted MCP dry_run=true or generation_admission_preview | Admission and cost preview only; the job is not submitted. | Safe automation |
Which routes need no API key?
Six public routes answer without a Sume API key: GET /v1/health, GET /v1/catalog, GET /v1/openapi.json, GET /v1/bgm/catalog, GET /v1/bgm/categories, and POST /v1/bgm/pick. Every other /v1 endpoint requires a key, and requests without one are rate-limited per client IP.
GET /v1/catalog is also where you confirm live prices for the calls below; see Sume API catalog. The three BGM routes are covered in Background music API.
Which requests end up costing nothing?
Requests that never run cost nothing. On a Format run, a create that fails with a 4xx, an idempotent 200 replay, and a skipped run cost nothing, and so does a schema rejected with output_schema_invalid, because nothing runs.
Failed jobs and failed queue admission release or refund the reservation where applicable, while generation a Format run finished before a cancel or a failure is billed. Do failed AI video generations cost credits? covers refunds, cancels, and retries.
Which calls charge one amount per call?
These tools price each accepted call as one unit rather than by length or size. Paid calls draw from one wallet at the published rate plus a 5.5% agent fee by default.
| Call | Basis | Price |
|---|---|---|
Create an avatar: POST /v1/avatar-1.0/generate | Per avatar, one time | $0.95 per avatar |
Music Router: POST /v1/music-router/generate | Per audio generation, every model | $0.125 per audio generation |
Remove background: POST /v1/rmbg-1.0/remove | Per image | $0.0225 per image |
Image upscale: POST /v1/image-upscale-1.0/upscale | Per image | $0.20 per image |
Trending video search: POST /v1/trending-videos/search | Per accepted call; summary_mode: metadata included | Listed in GET /v1/catalog |
What do the other paid calls charge?
The rest reserve a ceiling per job, or meter per unit of input or output:
- Standalone captions (videos up to 60 seconds) and the media tools, trim, filter, and detach and Timeline compose and audio, reserve a per-job amount listed in
GET /v1/catalogand never capture more than that. The video filter check stays free. - A Timeline render reserves $0.10 per output minute, counted per started minute of the audio spine, and never captures more than that reservation.
- Speech-to-text is $0.01 per audio minute, and a video inspect transcript bills at the same STT rate.
- Text to speech is $0.0475 per 1,000 characters; video upscale is $0.009 per video second.
- Routed image and video models are priced at provider list rates times Sume's margin, with per-model rates in the catalog.
What do the free checks not guarantee?
A passing check lowers risk; it does not promise the paid call succeeds:
- A video filter program that passes
/checkcan still fail on the box, from a bad expression, memory, or time, as a structured job error. - The Timeline plan cannot predict short-source pad or loop warnings.
- Neither check needs an
Idempotency-Key, but the paid render or encode that follows does.
Sources
Related posts
More in Pricing
- What is the Sume agent fee? The default 5.5%, explained
The Sume agent fee is charged on top of a request's model amount, 5.5% by default. Where the API reports it, and how to read the fee line.
- AI avatar video API pricing: cost per second and per minute
Sume bills AI avatar video per second by quality tier, with separate rates when you send a product image. Per-minute costs for standard, plus, and max.
- AI video generation cost per video: what one Sume API run cost
To see what one Sume run or video job cost, call GET /v1/usage with run_id or job_id and read debited_usd: the wallet deduction, agent turns included.
- Insufficient credits error 402 on the Sume API: add funds and retry
A Sume 402 insufficient_credits means the balance cannot cover the request's estimated cost, so nothing ran. Check GET /v1/balance, add funds, then retry.
Written by Sume