Blog
Sume blog: agent-native video generation, Avatar 1.0 orchestration, and production APIs for builders and coding agents.
OpenRouter's video expired event has no Sume twin: one normalizer
OpenRouter sends completed, failed, cancelled and expired video events; Sume sends three job events. A Python normalizer and verifier for both.
One length gate for six ad platforms: caps table and Sume plan
Reels 15 min, Stories 60 min, TikTok 10 min, LinkedIn 30 min, Pinterest 5 min, Snap 180 s. Check one Sume Timeline length against all six with the plan call.
Twenty image jobs, one webhook: mode webhook plus a Python verifier
Submit 20 image requests with mode webhook, receive signed job.completed callbacks, and verify them in Python. Retries, replay window and the poll fallback.
AI music for Reels, TikTok and Shorts: what Sume's Terms say
Can you post a Sume Music track on Reels, TikTok or Shorts? Sume's Terms claim no ownership and paid plans include commercial use. Platform rules are separate.
WAND-Vega on Sume: price ladder from Lite to Pro 1.5
Sume's Tencent-only WAND-Vega rows bill from $0.072 to $1.35 per second. Full ladder for Lite, Pro 1.0 and Pro 1.5 across 720p to 4K, text-to-video, 4 to 15 s.
Wan 3.0 edit and extension at Alibaba vs Sume's Wan row
Alibaba lists Wan 3.0 editing and extension; Sume's wan-3.0 row does text, image, end frame and references only. Edit with Omni Flash 1.1 or H3 Max Recast.
Vidu Q4 has no text-to-video: Sume rows that take a plain prompt
Vidu's Q4 page lists only image-to-video and reference-to-video. Sume has no Vidu row, but several rows start from a text prompt alone. Compare limits.
Text-only hero art: Imagen 4 Ultra or Nano Banana 2.1 on Sume
Imagen 4 Ultra costs 0.075 USD, takes no references and lists 5 ratios. Nano Banana 2.1 costs 0.10 and lists 15. Which to pick for 100 hero images.
Zapier Delay After Queue: space out Sume submits
Zapier's Delay After Queue releases held tasks one at a time, so a burst of rows does not hit Sume together. Here is the setup, the limits and the cap to set.
Slack scheduleMessage: announce a finished Sume render
Slack lets a bot schedule a post up to 120 days ahead, with at most 30 per 5 minutes in a channel. Finish the Sume render first, then schedule the announcement.
Format works in chat but fails over the API? The approval gate
A Sume Format that pauses for approval in chat runs unattended over the API. See what changes, why unattended_blocked appears, and a pre-flight list.
Vercel protection bypass in a Sume webhook URL
Sume cannot add a custom header to a webhook, but Vercel accepts the bypass secret as a query parameter. Check what Sume keeps and who can see the URL.
Know the day Sume lists a new video model: a catalog check script
Vendors launch video models weekly. A short Python script reads GET /v1/videos/models and flags when an id or name appears, with exit codes for cron or CI.
Notion X-Notion-Signature vs Sume sume-v1: two verifiers
Notion signs the body with your verification token. Sume signs a timestamp plus the body. Here is how to keep the two checks apart in one receiver.
Apps Script fetchAll: check 20 Sume run statuses in one call
fetchAll takes an array of requests and returns an array of responses. Use it to poll many Sume run statuses in one trigger without a loop of fetches.
Vidu Q4 voice references vs Sume's reference audio inputs
Vidu Q4 takes up to 3 voice references. Sume has no Vidu row; Seedance, Wan 3.0 and MiniMax H3 accept audio references, but voice cloning is app-only, not API.
Meta skips AI labels: no significant edit, no real-looking human
Meta applies no AI label when its own tools make no significant edit and no photorealistic human. See how Sume trim, crop and dim relate, from Meta's text.
Avatar video request rejected? Pre-flight these 6 inputs
Catch Sume Avatar 1.0 rejections before you submit: script vs video_inputs, the 4-60 second window, silence scenes, HTTPS media and caption style. A checklist.
Does your Format webhook receiver pass OWASP's checklist?
Check a Sume Format webhook receiver against OWASP's webhook cheat sheet: raw body, HMAC, five-minute window, dedupe on request_id, and where docs are silent.
Grok Imagine Lite upscales to 1080p: what Sume offers instead
xAI describes Grok Imagine Video 1.5 Lite as lowest cost with upscaled 1080p. Sume has no Lite row; it lists grok-imagine-video-1.5 at a flat rate. Compare.
Kling 3 on Sume: text or start and end frames, no reference inputs
Sume's kling-3 row takes text or start and end frames, 4 to 15 s at 1080p, with no reference images, video or audio. Prices and rows to use instead.
List Sume's music and TTS router engines in Python before deploy
A short Python script reads the music and TTS router model lists so a deploy check can confirm your pinned engine ids and fixed prices still exist.
Grok Imagine ignores aspect_ratio on image-to-video; Sume rejects it
xAI says image-to-video output matches the input image and ignores aspect_ratio. Sume's Grok row goes further and rejects the field. Crop the still first.
YouTube AI disclosure for AI music and voiceover: what to keep
YouTube asks creators to disclose meaningfully altered or synthetic realistic content. What to keep from Sume music and TTS jobs, and what Sume cannot decide.
TTS output_format: choose wav or mp3 for joins, captions and clips
Sume TTS returns mp3 by default. Choose wav when you will join, slice per sentence or feed lip-sync; mp3 is smaller but adds padding at every edge.
Creatomate template modifications vs a Sume Format run input
Creatomate fills a fixed template through modifications; a Sume Format run takes free-form input and generates new media. Request shapes and when each fits.
YouTube's four disclosure triggers checked against Sume's tools
Four YouTube disclosure triggers: real person, altered real footage, invented realistic scene, music. Each is checked against Sume generation and media tools.
Thrift and resale listing videos from one flat-lay photo
A reseller can turn a flat-lay photo into a 6-second listing clip with Wan 3.0 first-frame video for $0.38 at 480p. Rules for honest motion and a batch run.
Mingmou 1 and Hunyuan 1.5 on Sume: fixed lengths and low rates
Sume's Tencent rows Mingmou 1 and Hunyuan 1.5 bill $0.05775 per second at 720p, take 5 or 10 seconds (Hunyuan 5 only), no audio. What they can and cannot do.
YouTube's resolution ladder: which rows Sume Timeline can render
YouTube lists eight 16:9 sizes from 426x240 to 7680x4320. Sume Timeline renders even edges from 256 to 2160, so 1080p, 720p, 480p and 360p fit; the rest do not.
Detach 13 seconds of speech from a clip and lip-sync it on H3 Max
Sume can detach audio from a hosted clip by time range, then drive a talking face with H3 Max lip-sync. $1.31 for 13 seconds; rights and limits explained.
Gradium TTS covers 5 languages: what Sume's language field does
Gradium lists English, French, Spanish, German and Portuguese. Sume TTS takes a language field and a mismatch guard; here is how a non-English line behaves.
Live-sale replay to 10 highlight clips: import, trim, caption
Cut ten highlight clips from one TikTok live replay with Sume: $0.15 import, ten $0.02 trims, ten $0.20 captions. Total $2.35. Limits stated plainly.
Send product photos to the Sume agent API, get edited images back
Agent Completions takes up to 30 image attachments and a required spend cap, then returns generated images in output.images. When to use it over /v1/images.
Gradium's 329 character voices vs voices of Sume avatars
Gradium lists 329 character voices. On Sume, a spoken character is an avatar with a ready voice; Avatar 1.0 is English-only. What that means for casting.
YouTube: upscale and repair need no AI label, realistic edits do
YouTube's page exempts sharpening, upscaling or repair, beauty filters and colour changes from AI disclosure. Realistic synthetic people or places need it.
Port a fal queue submit and poll loop to Sume /v1/videos in Python
A fal queue client posts to queue.fal.run, polls a status URL and fetches a result. The Sume version is 21 lines of Python on /v1/videos. Field map, traps.
GitHub environment reviewers: approve before a paid Sume run
Put the Sume key in a GitHub environment with required reviewers, so a workflow pauses for approval before it starts a paid run, then cap the run's spend.
Bike shop Black Friday clip: dim the footage so price text reads
A bike shop can darken a generated clip with Sume video filter dim, then burn price cues on it for $0.20. One 5 s clip at 720p is $0.85.
TikTok ad 516 kbps bitrate floor: smallest file for 10 minutes
TikTok lists a minimum bitrate of 516 kbps and a 500 MB cap for non-Spark ads. The math gives 38.7 MB at the floor for 10 minutes, and Sume cannot tune bitrate.
WordPress Action Scheduler: poll a Sume run as a queued action
Action Scheduler claims 25 actions per batch and marks any running over 5 minutes as failed. Schedule a short Sume poll action instead of waiting inside one.
Grok Image on Sume: one image per call, so fan out four in Python
x-ai/grok-image lists n as 1 to 1 in the Sume catalog. How to get four variants with four parallel calls, what it costs, and how to stay under queue limits.
YouTube AI label: in the player or the description?
YouTube shows its altered-content label in the player for photorealistic AI and in the expanded description for animated or non-photorealistic AI.
Pinterest video Pins: crop a 9:16 master to 1:2, 2:3, 3:4, 4:5
Pinterest lists 9:16 full-bleed plus 1:2, 2:3, 3:4, 4:5 and 1:1, from 4 s to 5 min. Centred crop fractions from a 9:16 source with Sume video-filter.
OpenRouter: poll video every 30 s. Sume: obey next_poll_after_seconds
OpenRouter recommends a 30 second poll on video. Sume's jobs API returns next_poll_after_seconds, else backoff. A Python poller that does both.
Pressure-wash and gutter reminder: one clip, six neighborhood captions
Burn a different neighborhood name into one silent clip with Sume video captions cues, $0.20 per version. Six versions cost $1.20 on top of the base clip.
Descript-style delete-a-word editing with Sume STT and video trim
Descript edits video by editing its transcript. Sume has no such editor, but STT word timings, video trim at $0.02 a cut and Timeline can rebuild the result.
n8n Loop Over Items batch size vs Sume bulk concurrency 16
Loop Over Items sends one batch at a time. Send each batch to one Sume bulk run, with up to 100 items and concurrency 1 to 16, not one request per item.