Blog
Sume blog: agent-native video generation, Avatar 1.0 orchestration, and production APIs for builders and coding agents.
Sume schedule run: communication.mode webhook alone sends nothing
On a Sume schedule run, communication.mode is descriptive only. Delivery starts when communication.webhook_url is set, an HTTPS URL up to 2,048 characters.
Sume 415 unsupported_media_type: read details.received_content_type
A 415 from the Sume API means the body was not sent as application/json. Which client defaults cause it, how to read the details field, and the one-line fix.
Fresh Idempotency-Key per proxy call: why a Sume retry bills twice
If your server proxy mints a new Idempotency-Key on every request, a browser retry becomes a second paid Sume job. Forward the client's key instead; TypeScript.
X website-button video: headline sits 8-12 px in, move captions
X's spec puts the headline 8 to 12 pixels from the left and bottom of a website-button video. Keep burned-in Sume captions clear of that corner at 800x450.
Shortest valid clip on each Sume video model and what it costs
The minimum duration differs by model: Wan 3.0 starts at 2 s, Omni Flash at 3 s, Seedance at 4 s, MiniMax H3 at 5 s. Cost of one minimum clip at two sizes each.
X standalone video lists 16:9 and 1:1: set the Sume render size
Sume timeline renders 1080x1920 by default, but X's standalone video ad spec lists 16:9 and 1:1. Set output width and height to 1920x1080 or 1200x1200.
Pinterest video ads start at 4 seconds, Sume trim allows 0.2: gate it
Pinterest's standard video ad minimum is 4 seconds and 15 minutes max, but video trim accepts 0.2 to 900 seconds. Add a length gate before you upload.
Korean ad with an English brand name: set Sume TTS language to ko
Send a Korean script with language set to ko; the default is English. A brand name in Latin letters is read by the chosen language, so listen and fix it.
Size a Sume submit wave: limit 100, 30 processing, 10 queued gives 60
How many new jobs can a Sume workspace take? max(0, limit - active - queued), capped by queue_capacity_remaining. Worked numbers and a TypeScript helper.
TTS speed 1.2 turns a 30-second read into about 25 seconds
Sume TTS accepts generation_config.speed from 0.6 to 1.5. Speed changes length, not the bill: 450 characters cost $0.021375 at any speed. Test the real length.
Nano Banana Pro costs 1.25x the 2.1 at 2K on Sume, 4x at Google 1K
Google lists Pro at $0.134 and 2.1 at $0.0336 for 1K (about 4x). Sume bills Pro $0.1875 and 2.1 $0.15 at 2K, only 1.25x. The ratio at each size.
Does a lip-synced dub in your own voice need YouTube's AI label?
YouTube says cloning your own voice for voice-overs or dubs needs no disclosure; a real person made to say what they never said does. How Sume lip sync fits.
Runway Gen-4.5 vs Luma Ray3.2: price of a 5 and 10 second clip
Runway lists Gen-4.5 at 12 credits a second ($0.12). Luma lists Ray3.2 per block: $0.30 for 5 s at 720p and $0.90 for 10 s. Here is the full grid.
One frame at 12.5 s: video-frames vs video-inspect
Video frames returns a frame at its source size from clips up to 300 s. Video inspect defaults to 768 px, caps at 2160, and takes clips up to 1,800 s.
Vidu: failed and moderated videos use no credits. What Sume does
Vidu's pricing page says failed generations, moderation rejections included, use no credits. Sume reserves at submit and releases on failure. How to check each.
Python cost cap for a mixed media job: round up, then refuse
A 25-line Python estimator with Decimal prices that rounds the total up to the cent, as Sume's reservation does, and exits before submit if it is over your cap.
X video ad under 30 MB: how many seconds fit at 6,000 and 5,000 kbps
X says keep video ads under 30 MB. At its 6,000 kbps (1080p) and 5,000 kbps (720p) targets that is 40 and 48 seconds. Check yours with Sume video inspect.
Shorts library track limited to 30 or 60 s: share of 3 minutes
YouTube says most Shorts songs run 90 seconds in a 3-minute Short, some 60 or 30: 50, 33 or 17 percent. See how a Sume soundtrack bed fills the rest.
Runway lists Wan 3 and Seedance 2: 10-second price vs Sume
Runway's API sells wan3 at 10 credits a second at 720p and seedance2 at 36. Here is the 10-second price at each resolution next to Sume's list rows.
Face swap API sync mode waits at most 30 s: then poll, do not resubmit
Sume's job modes cap sync and subscribe waits at 30 seconds. A face swap that is not terminal by then returns the envelope: poll the job, never resubmit.
16:9 clip in Snapchat's 720x1280 frame: contain fills 31.6 percent
A 1280x720 clip placed in a 720x1280 Snapchat frame fills 405 of 1,280 rows with contain. See what cover crops and what blur does in a Sume timeline render.
Video inspect on a 60 s clip: stills at 3.75 s, then every 7.5 s
With no frames program, video inspect returns 8 mid-bin stills. On a 60-second clip they land at 3.75, 11.25, 18.75 seconds and on, 768 px on the long edge.
Claude Code: local Sume entry vs claude.ai connector, who wins
Claude Code ranks a local Sume MCP entry above a claude.ai connector. That decides whether you get the API key's full tool set or OAuth's read-only one.
Paying for seconds you cut: generate 30 s, keep 12, on Sume video
Sume bills the seconds you generate, not the seconds you keep. Asking Wan 3.0 for 30 s and using 12 costs $3.75 for 12 s of use, $0.3125 per kept second.
A 5-minute voice-over: 4.8 MB as 128 kbps MP3, 26.5 MB as WAV
Sume TTS defaults to MP3 at 44,100 Hz and 128 kbps. Five minutes is 4.8 MB; 16-bit mono WAV is 26.46 MB at 44.1 kHz and 28.8 MB at 48 kHz.
Snapchat Commercials: 3-6 s or 7-180 s, a 6.5 s cut fits neither
Snapchat lists Standard Commercials at 3 to 6 seconds and Extended Play at 7 to 180. Cut to the exact second with video trim exact and check duration_seconds.
Switching Fabric to H3 Max Lip Sync: 6 s, 12 s and 20 s clips compared
Fabric and H3 Max Lip Sync share a body on Sume, so switching is a URL change. At 6 s H3 768p is $0.60 vs Fabric 720p $1.125; at 20 s H3 Max is not eligible.
TikTok AI disclaimer sits at the bottom: keep burned captions higher
TikTok's ad disclaimer shows at the bottom of in-feed video for the whole ad. Place Sume burned-in captions higher with placement.anchor_ratio and test.
A 300-second recording in 24 stills: video-frames fps 0.08
Video frames caps a call at 24 stills and a 300-second source. Send fps 0.08 and a 300-second clip gives 24 mid-bin stills, one every 12.5 seconds.
Ray 3.2 extend bills one 5-second block whatever duration you send
Ray 3.2 single-keyframe extend is $0.15, $0.30 or $1.20 per call at 540p, 720p, 1080p, even if duration is set. Chain math and Sume's per-clip equivalents.
AI video API billing units: per clip, per second, credits or tokens
Luma bills per generation, LTX per second, Runway and Vidu in credits, Google in tokens, MiniMax per second plus inputs, Sume in USD. Worked examples.
7-second 720p AI video clip: Luma lists 5 or 10 seconds, Sume bills 7
A 7-second 720p clip: $0.56 on MiniMax, $0.63 on LTX Fast, $0.525 on Sume H3. Luma's page lists only 5 and 10 second totals ($0.30, $0.90).
LTX-2.5 Pro costs 30-33% more than Fast at every resolution
On LTX's price page Pro is $0.12-$0.39 per second against Fast at $0.09-$0.30. That is +30 to +33% at each tier; per-minute math and the nearest Sume rows.
Vidu Q4 Preview 4K costs 8.7x the 540p price: 78 credits vs 9
Vidu's table runs 9 to 78 credits per second, 540p to 4K. The two published USD points ($0.045, $0.39) imply half a cent per credit. Plus Sume's 4K row.
Vidu 500 RMB top-up: 16,000 credits is 1,777 seconds of Q4 at 540p
Vidu's smallest API package is 500 RMB for 16,000 credits. How many Q4 Preview seconds that buys at each resolution, and the flat 32 credits per RMB.
waitForRun or waitForJob? Pick by which Sume endpoint made the id
A Sume run id cannot be read as a job id. Which create endpoint made your id decides waitForRun or waitForJob, with a TypeScript example of each.
Do Sume result URLs expire? media.sume.com vs /videos content
Sume Format artifact URLs on media.sume.com do not expire and are public. The /v1/videos content route needs your key. What to store, plus a Python download.
Sume job next_action: poll_status, fetch_result, inspect_events
The Sume status payload tells your client what to do next. Three next_action values map to five job states; here is the full table and a TypeScript switch.
Sume /v1/videos says cancelled, /v1/jobs says canceled: a status map
Two spellings and two vocabularies for the same Sume job: pending to cancelled on /v1/videos, queued to canceled on /v1/jobs. A table and a TypeScript mapping.
Move an OpenRouter-style video client to Sume: the 400s to expect
Sume /v1/videos matches the OpenRouter video shape, but size, provider.options and seed return 400. The differences table and a Python asyncio polling client.
A Sume job failed: read GET /v1/jobs/:id/events, then act
The events endpoint is the first stop for a failed or canceled Sume job: the eight event names, what stays hidden, error categories, and a TypeScript reader.
Hy Image 3.5 Preview: 17.44 s median latency vs Sume's 30 s wait
OpenRouter shows a 17.44 s median for Hy Image 3.5 Preview. Sume lists no Hy row, but its image calls wait 30 s, then return a 202 job. Handle both.
Hy Image 3.5 Preview at 91.57% availability: retry math for 200 images
OpenRouter shows 91.57% availability for Hy Image 3.5 Preview. For 200 images that is about 17 failed first tries. The math, and Sume's failed-job billing.
One provider behind Hy Image 3.5 Preview vs Sume's one image endpoint
OpenRouter forwards Hy Image 3.5 Preview to one provider, Tencent Cloud. Sume's image API serves every model from one endpoint. What each means for routing.
MAI-Voice-2.1 voice cloning is gated; how Sume TTS picks a voice
MAI-Voice-2.1 clones a voice from a 5 to 60 second clip but needs gated access. Sume TTS selects voices by avatar or voice id and has no reference-audio field.
FLUX.3 Image vs Seedream 5 Lite: price per image at 1K, 2K, 4K
BFL lists FLUX.3 Image at $0.048 (1k), $0.100 (2k) and $0.607 (4k). Sume lists Seedream 5 Lite at $0.04375 and no FLUX.3. Full grid and 1,000-image totals.
FLUX.3 Image costs 4.1 to 60.7 credits: what 1,000 credits buys
BFL prices FLUX.3 Image at 4.1, 4.8, 10 or 60.7 credits (1 credit = $0.01). 1,000 credits is 243, 208, 100 or 16 images. Sume does not list FLUX.3.
Nano Banana 2.1 4K is 3,780 tokens, 3.4x the 1K cost: Google vs Sume
Google prices Nano Banana 2.1 by image tokens: 1,120 at 1K, 1,680 at 2K, 3,780 at 4K. Sume bills a flat tier price instead. Both price ladders, side by side.