Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
How do I caption a Shorts series in two languages in one batch?
Run one Sume caption job per episode and language: $0.20 each for clips up to 60 seconds, so a 10-episode season in two languages is $4.00.
How do I make a 180-second Shorts episode from 30-second Wan clips?
A 180-second episode is six 30-second Wan 3.0 clips joined by one Timeline render: $22.50 for video at 720p plus $0.30 for the render, in Sume prices.
How do I reuse one theme track across every Shorts episode?
Detach or build the audio once, then join parts with Timeline audio concat: $0.01 per job, up to 20 ordered parts in one gapless file.
What does a 10-episode Shorts season cost on Wan 3.0 at 720p?
Ten 60-second episodes on Wan 3.0 cost $75.00 at 720p, $37.50 at 480p and $150.00 at 1080p in Sume list prices. Per-second rates and the math inside.
Can Sume import my existing YouTube Shorts to reuse in a new episode?
No. Sume's media import takes TikTok and Instagram links at $0.15 and rejects YouTube links. For your own Shorts, work from the original master.
Fix one detail in a Shorts episode: Omni Flash edit or re-roll?
Gemini Omni Flash 1.1 can edit an existing clip through video_url instead of regenerating it. A fresh 10-second re-roll at 720p costs $1.25 on Sume.
Redo one scene of a Shorts episode without redoing the episode
Continue the same Format thread with previous_run_id and a new instruction to redo a single scene. Sume treats it as one more turn of the same conversation.
Will polling video jobs trigger a 429? Reads and writes differ
Each Sume API key has a write budget and a read budget 40 times larger, so a status-poll loop cannot 429 your submits. Poll math, headers and a 429 backoff.
Retry a timed-out video submit without paying twice: Idempotency-Key
Your client timed out after submitting a 30-second video job? Retry with the same Idempotency-Key and Sume returns the original job, not a second bill.
Seedance 2.5 reference limits: 30 images on the vendor page, 9 on Sume
ByteDance lists up to 30 images, 10 video and 10 audio references for Seedance 2.5. Sume's Video Router validates fewer: 9, 3, 3, and 12 in total. Read both.
Verify a Sume webhook in Python when the secret is rotating
A Sume webhook can carry two sume-v1 signatures for 24 hours after rotation. A FastAPI verifier that accepts either, rejects an empty secret, checks the clock.
Wan 3.0 reference-to-video body: images, videos and audio in one job
A Wan 3.0 reference job on Sume takes up to 10 images, 5 videos and 5 audio files. Request body, the audio rule, and the frame-versus-reference conflict.
Poll or webhook for a 30-second AI video job? Use both
A 30-second Seedance 2.5 or Wan 3.0 clip runs for minutes. Submit async, take the signed webhook, and keep one status poll as the backup. Code included.
Sync mode returned queued on a 30-second video: the 30 s cap
mode sync waits at most 30 seconds on the HTTP call, not on the job. A 30-second clip returns a job id with sync.timed_out true; keep polling, never resubmit.
How many 30-second jobs can you submit at once? Read generation_limits
Submit responses carry generation_limits. Use concurrency_limit minus active and queued jobs as your in-flight budget, and treat wave_size_hint as a hint only.
Verify Sume job webhooks in Express: raw body, sume-v1, replay window
Verify a Sume job.completed webhook in Express with @sume-com/sdk verifyWebhook: raw body, empty secret refused, dedupe on job_id, answer 2xx fast.
What a 30-second Wan 3.0 job reserves on Sume: list price x 1.25
At submit Sume reserves provider list x 1.25. For Wan 3.0 at $0.05, $0.10, $0.20 per second, a 30-second job reserves $1.88, $3.75 or $7.50. Handle 402.
waitForJob for a 30-second Seedance clip: TypeScript SDK example
With @sume-com/sdk, submit a seedance-2.5 job through generateVideoRouter and let waitForJob poll it for up to 20 minutes. Idempotency key, onStatus, errors.
Seedance 2.5 30-second job: set a 20-minute client deadline in Python
A 30-second Seedance 2.5 job outlives any HTTP timeout. Submit async to Sume, poll with next_poll_after_seconds, and set the deadline in your client.
Lost a 30-second video job after SSH died? Sume CLI recovery
There is no sume video subcommand, but the Sume CLI recovers any job you made through the API: jobs status, watch, result and download. Never resubmit.
Caption job failed: caption_no_speech on a music-only AI video
Sume's caption job fails with caption_no_speech when a clip has no audible speech. Send cues with text, start and end instead; it is the same $0.20.
TikTok x Sundance micro-series course: what a pilot costs to render
TikTok and Sundance Collab run a free four-week micro-series writing course. Here is a 60-second pilot priced as ten Wan 3.0 shots on Sume: $7.80.
Recast the actor in a short video with AI: H3 Max Recast, TikTok rules
H3 Max Recast swaps the people in a 5-30 s source video for 1-4 photos you supply. TikTok's Shop pages ban impersonating public figures and need AI disclosure.
Fix a wrong product color in an AI clip with Gemini Omni Flash edit
TikTok Shop bans AI that changes a product's color, size or material. Gemini Omni Flash 1.1 on Sume edits one clip from a prompt instead of re-rolling it.
TikTok Shop UK bans high-volume repetitive AI content: vary variants
TikTok Shop's UK AI page lists high-volume repetitive, low-effort content as prohibited. Make each variant differ in hook, shot and script, not in a filter.
Review 30 episode exports fast: video-inspect seek fast vs precise
Sume video-inspect gives 8 stills by default. Seek fast jumps to a keyframe at or before each time; precise lands on the frame. Pick by the check.
Duck music under dialogue and fade out a TikTok mini drama episode
Sume's Timeline render takes a soundtrack with duck_db (0-20) and a fade_out_seconds up to 5, so dialogue stays clear and each episode ends on a clean fade.
Get a transcript of an AI video before TikTok Shop review
Run Sume video inspect with transcribe: true to read what an AI video says, at $0.01 per audio minute plus compute, before you post a TikTok Shop clip.
TikTok auto captions or burned-in captions on an AI video?
Use both when it matters: TikTok's auto captions are a viewer toggle that needs speech; burned-in captions from Sume ($0.20 per clip up to 60 s) always show.
Chain TikTok mini-drama episodes: last frame in, first frame out
Pull the final still of episode one with Sume video frames, then send it as first_frame for episode two so the next shot starts where the last ended.
TikTok Shop UK bans fear-based organ imagery: lint your AI prompt
The UK Shop AI page bans fear-based health imagery like diseased organs. Check your prompt before rendering, then check frames with Sume video-frames.
Wan 3.0 draft at 480p, final at 1080p: no seed means a full re-render
Sume lists Wan 3.0 at $0.0625, $0.125 and $0.25 per second at 480p, 720p, 1080p. No v1 model takes a seed, so a draft is a prompt test, not a preview.
Do fictional AI actors in a TikTok mini drama need the AI label?
Yes if the faces or voices are realistic and AI-made: TikTok asks creators to label content fully generated or significantly edited by AI. Here is how.
Business terms pronounced right: Decagon tuning vs fixing a TTS script
Decagon lets teams tune pronunciation of business terms in Voice 3. For a voiceover job you fix it in the script. A practical test loop and its cost.
What is Decagon Chord? A speech model for support calls, not ad reads
Chord is Decagon Labs' voice model behind Voice 3, launched Oct 1, 2026. What the announcement says, and when you need a TTS job instead.
MAI-Voice-2.1 or Flash for narration? Microsoft's own guidance
Microsoft lists MAI-Voice-2.1 for long-form narration and Flash for agents. The prices, the latency claims and which one to test first for voiceover.
Licensed data, consented voice talent: 6 questions for your TTS vendor
Decagon says Chord trained on licensed data and consented voice talent. Six questions to ask any text-to-speech vendor before you ship an AI voice.
Phone numbers and order codes in an AI voiceover, kept clear
Decagon says Chord slows down for confirmation codes and phone numbers. How to get the same clarity in a script-based voiceover by writing digits well.
Decagon Voice 3 price: what is public and how to size a pilot
Decagon's Voice 3 launch post lists no price. How to size a voice-agent pilot in call minutes, and what the speech parts cost on public rate cards.
Decagon Voice 3 duplex agent or a voiceover job: which do you need?
Voice 3 listens while it speaks. A voiceover job does not. How to choose between a live voice agent and a file-based TTS job for your project.
What does one minute of AI speech cost? Gemini, MAI, Voxtral and Sume
Convert per-token, per-character and per-minute TTS prices to a cost per finished minute, with a stated 900 characters per minute assumption.
Voice agent, TTS API or rendered narration? Pick by output
Decagon Voice 3, MAI-Voice-2.1 Flash, Gemini 3.8 TTS and Sume jobs solve different problems. A decision table for choosing by output, October 2026.
Decagon says Chord is 90% indistinguishable: run your own blind test
Decagon cites about 90% indistinguishability across three voice samples. How to run a fair blind listening test on any TTS output, with a cost estimate.
Let a video agent read before it spends: MCP OAuth read, then write
Hosted MCP OAuth gives mcp:read by default and mcp:write by a consent toggle. Let an agent plan with reads, then enable write for paid calls.
Format vanity URL or skl_ invoke_url: store the opaque path
A Format has a vanity URL (handle/slug) and an opaque skl_ invoke_url. Renamed handles keep resolving for 90 days, but only the skl_ path survives every rename.
One bad item in a 100-item Format bulk request: 400, nothing runs
A Format bulk create validates every item first. One invalid item returns 400 with details.index and queues nothing.
Next-episode end card in Timeline 1.0: held still, fade, cliffhanger
Add a next-episode end card to a micro-drama with a Timeline 1.0 still slot: stills hold static, fades run up to 1 s, render is $0.10 per output minute.
Recurring AI lead in a Shorts series: label every episode?
YouTube asks for disclosure on realistic generated content; TikTok asks for labels on fully AI-generated or heavily edited content. Record each episode.