Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
Can a YouTube Shorts episode run 4 minutes? Split with source_in
No: the Shorts series limit is 180 seconds. Split a long cut into two parts with Sume Timeline audio.source_in and video source_in, one render per part.
Kling 4.0 or Seedance 2.5 for a 30-second AI video today?
Both advertise 30 seconds. Only one has a callable id on Sume today. A spec-by-spec read of the vendor pages and what you can actually request.
Lyria 3.5 better vocals and pronunciation: a listening checklist
Google says Lyria 3.5 improves vocals, lyrics and pronunciation. Check a Sume Music 1.0 take yourself: listen, transcribe it with STT, and compare the words.
Pick a score from three Lyria takes: cost and idempotency keys
Three music takes on Sume cost three generations. Use one idempotency key per take, name the differences in the prompt, and keep the pick for the timeline.
Transcribe a Zoom recording to text with an API, in 10-minute chunks
Turn a long meeting recording into text with Sume: detach the audio from the MP4, cut it into ranges under 10 minutes, and send each range to STT 1.0.
Omni Flash or Veo 3.1 for the Gemini API? What Sume runs
Google's video docs now call Omni Flash the default over Veo 3.1. Here are Google's listed prices for both, and which one Sume's video router carries.
YouTube show episode numbers follow publish date: render in that order
YouTube assigns Shorts episode numbers by publish date unless you set a manual playlist order. Plan your render queue so numbers match your story.
Black Friday grocery and toy promo videos with AI: where to start
Adobe's forecast has grocery up 10.3% and toys up 9.6%, the fastest of its categories. Which Sume Formats fit a food or toy promo, and what to check.
AI customer service with a video avatar: where a clip fits
Tavus compared eight AI customer service platforms on 2026-10-02. A live agent answers the long tail; a rendered avatar clip covers top answers. Costs.
What is a consistent story for a YouTube show? AI series checklist
YouTube asks shows for the same characters or hosts, one long story, or one topic. A checklist for keeping an AI-made series consistent across episodes.
AI video upscaler API fields: factor 1.1 to 4 and fast, standard, pro
The Sume Video Upscale request takes upscale_factor from 1.1 to 4, duration_seconds from 1 to 30 and an enhancement_tier. What they do and cost.
Cost of 1,000 AI images on Sume: Soul, Flux 2 Pro, Grok, Nano Banana 2
A thousand single-image jobs cost $10 on Higgsfield Soul, $40 on Flux 2 Pro, $30 on Grok Image and $100 on Nano Banana 2, from Sume's catalog prices.
Will a YouTube show accept audio-led Shorts with no real visuals?
YouTube says videos with no visuals likely will not be considered a show. How to give narrated, audio-led Shorts a real picture track before you build a series.
YouTube watch series button: viewers may land on episode 7 first
A viewer who finds one Short in their feed sees a watch series button, so any episode can be the first they see. Write and render episodes that stand alone.
Runway free plan 125 credits: how many seconds of Gen-4.5 is that
Runway's free plan gives 125 one-time credits and Gen-4.5 costs 60 credits per 5 seconds. That is 10 seconds. What the same 10 seconds cost on Sume.
Interactive AI video for enterprise: live avatar or quiz video?
Interactive can mean a live conversation or a quiz inside a video. A table of both from HeyGen's September guide, and which part a Sume avatar clip covers.
AI music video generator API: which models take your audio?
Seedance, Wan and MiniMax accept audio references on Sume. Gemini Omni and Kling 3 do not. Limits per model for a music-video workflow.
Turn a voice memo into a narrated short: transcribe, edit, re-voice
Speak your idea into a phone, transcribe it with Sume STT, tidy the text, then re-voice and caption it. The memo becomes the script, not the audio.
Build a Sume Idempotency-Key from an order id: changed body result
The same key and body replays the original Sume job. A different body with the same key is a 409. Pick keys that make both outcomes safe, in runnable Python.
MAI-Voice-2.1 clones from seconds: how Sume clones from a clip
Microsoft says MAI-Voice-2.1 clones from a few seconds of audio with consent guardrails. In Sume Agents, voices_create clones from an https audio clip.
Check a Short has sound and picture before adding it to a YouTube show
YouTube asks shows for high-quality video and sound and says no-visuals videos likely will not count. Probe each Short with Sume video inspect first.
Does the Sume SDK retry a failed video submit? Only with a key
createSumeClient retries GET and HEAD by default and retries a POST only when an Idempotency-Key header is set. See what is retried, what is not, and why.
Ideogram is_image_safe in the response: what Sume returns instead
Ideogram's precise edit returns an is_image_safe field per image. Sume has no such field: a refused input ends as a failed job with generation_rejected.
Trim silence from a voiceover using STT word timings (Python)
Find the dead air in a voiceover from Sume STT words[] timings: list gaps over a threshold in Python, then cut with timeline audio ranges.
MiniMax H3 on Sume: 9 images, 3 videos, 3 audio, 12 in total check
On Sume, minimax-h3 and minimax-h3-max take at most 9 reference images, 3 videos and 3 audio clips, 12 in all, and audio cannot be alone. Check it in Python.
Thanksgiving to Cyber Monday ad plan with AI video Formats
Adobe puts Thanksgiving at $6.9B, Black Friday $12.9B and Cyber Monday $15.1B. A four-day content plan using Sume Formats, with a cap on each run.
Upload a reference image to Sume with curl: three requests, one URL
Get a durable HTTPS URL for a local image before a video call: reserve a presigned PUT, upload the bytes, complete the asset. A short curl and jq script.
Sume webhook beat your database write: handle an unknown job_id
A fast Sume job can send its webhook before your submit handler commits the job id. Park the event in an inbox table and reconcile it, in runnable Python.
Sume webhook signature mismatch? Check the secret fingerprint header
Each Sume delivery carries x-sume-webhook-secret-fingerprint. Compare it with the dashboard before debugging code. Python verifier that refuses an empty secret.
waitForJob threw SumeJobTimeoutError: resume polling with the job id
The Sume SDK stops waiting after 20 minutes by default, but the job keeps running. Catch SumeJobTimeoutError, read jobId and wait again. TypeScript example.
Captions app pricing is iOS-only: caption videos from code instead
The Captions pricing page says its prices reflect iOS plans only and names no API. Sume's caption endpoint burns captions for $0.20 a clip of up to 60 seconds.
Custom thumbnail for each Short in a YouTube series: pull the frame
YouTube lets a Shorts series carry a unique custom thumbnail per Short. Pull the best frame from each render with Sume video frames, then dress it.
Black Friday video ads for phones: which Sume Formats are vertical
Adobe expects 57.4% of holiday online sales on mobile. Which Sume catalog Formats are described as vertical, and how to ask for vertical in the rest.
Wan 3.0 lists 20 references; Sume caps 10 images, 5 videos, 5 audio
Wan 3.0's own page says up to 20 reference assets. Sume splits that into 10 images, 5 videos and 5 audio clips. Validate a request offline in Python first.
Runway Gen-4.5 cost per second by plan vs Sume per-second rates
Using Runway's annual prices, a Gen-4.5 second costs 23 cents on Standard, 15 on Pro and 9.6 on Max. Sume's per-second rates run from 6.25 to 25 cents.
YouTube show vs Shorts series: make long-form episodes with Timeline
A Shorts series must be Shorts only, up to 180 seconds. A regular show has no such cap on that page. Render long episodes, up to 30 minutes, with Sume Timeline.
YouTube preferred language setting: keep the source audio clean
YouTube now lets viewers set a preferred language for dubs. If you add your own dub, start from a clean speech track and detach it from the video first.
AI listing video disclosure line: put it in the Sume output schema
Carry the AI disclosure and the original-photos link with each listing video: a typed output_schema field, what input does not carry, and a check.
Gemini Omni 4K on Sume: send 4K or 4k, 10 s max, 16:9 or 9:16
Sume's gemini-omni-flash-1.1 takes 4K, and lowercase 4k works as an alias. Requests run 3 to 10 seconds, 16:9 or 9:16. Python body builder.
Does Sume STT detect the language? Omit language_code
Sume STT detects a clip's language when you omit language_code. MAI-Transcribe-2-Streaming claims continuous detection. They are not the same promise.
YouTube auto dub in 27 languages: how many can you voice on Sume?
YouTube auto dubbing covers 27 languages. Sume's voice library tags 16, and TTS needs a matching language field. Compare the two lists before you plan a dub.
Sora API shutdown alternatives: OpenAI lists no replacement
OpenAI's deprecations page removes the Videos API and sora-2 on 2026-09-24 and names no replacement. Pick a Sume video model by clip length and resolution.
YouTube show banner and backdrop: one has the title, one must not
YouTube wants a 16:9 banner with the show title and a 16:9 backdrop without it. How to make both from one source image with Sume Ideogram 4.5.
Serial or non-serial YouTube Shorts show: which should you pick?
YouTube lists non-serial episodes newest first and serial ones oldest first. Pick by whether a viewer can start anywhere, then render episodes in that order.
How many reference video seconds? Kling 4.0, Wan 3.0, H3, Omni
Reference video limits differ by model: Kling 4.0 allows 5 videos at 30 seconds combined, Omni 3 clips of 3 seconds. A table of the numbers I could read.
Replace the voice in a video you shot: detach, transcribe, re-voice
Swap the speech in a finished video on Sume: detach the audio, transcribe it, fix the script, speak it with TTS and render the picture with the new track.
Can a YouTube Shorts series episode be square? Render 1080x1080
Yes. YouTube's Shorts series rule says portrait or square, up to 180 seconds. How to render a square episode with Sume Timeline output width, height and fit.
Which API: speech, transcript, music, captions or audio track?
Pick the Sume endpoint by the job: TTS for narration, STT for a transcript, Music for a score, captions for burned text, audio detach for a video's sound.