Blog
Sume blog: agent-native video generation, Avatar 1.0 orchestration, and production APIs for builders and coding agents.
Shortest billable clip: Wan 3.0 at 2 seconds versus H3 at 5 seconds
Wan 3.0 starts at 2 seconds and MiniMax H3 at 5 on Sume. Shortest job: Wan 480p $0.13, H3 480p $0.38. Table of minimums for all models.
jq -n --arg: build a Sume video body when the prompt has quotes
Hand-quoted curl -d bodies break on a quote or backslash in the prompt. jq -n --arg escapes it right. Tested with a Wan 3.0 submit on curl 8.7.1 and jq.
Plant shop care-tip loop: Seedance 2.5 with one image as both frames
Send one photo as first_frame and last_frame to Seedance 2.5 on Sume for an 8-second 1:1 loop-style clip: the request body, $4.63 at 720p, and the limits.
curl -K config file: keep your Sume API key out of ps output
A key passed with -H shows in ps args and shell history. curl -K reads headers from a chmod 600 file instead. Tested on curl 8.7.1 against a stub.
Shorts split audio from video: which Sume models let you drop AI sound
The Shorts editor now separates audio and video tracks. Which Sume video models can generate silent footage, which always embed audio, and the price gap.
Grok Imagine Video on Sume: the cents-per-job ladder, 4 to 15 seconds
Grok Imagine Video 1.5 bills $0.0125 per second, rounded up to the cent per job. Every length from 4 to 15 seconds, and why a 5-second clip costs 7 cents.
MiniMax H3 clip audio: keep it or detach it for a Short
MiniMax H3 is reported to make native stereo audio. Detach it for $0.01 to mix it separately. Prices for H3 768p and H3 Max.
Shorts editor splits audio from video: make voice and music files
YouTube's Shorts editor now keeps audio apart from video. Make a clean voice file and a music bed on Sume for $0.163 and drop each on its own track.
Seedance 2.5 30-second clip as a whole Short: what it costs
One 30-second Seedance 2.5 clip at 720p is $17.334 on Sume. Compare with 8 and 15 seconds and with cheaper ways to fill 30 seconds.
Repost TikTok to Shorts: what a Sume rebuild chain costs
A TikTok or Reel you own, rebuilt as a Short with new voice, music bed and captions: six Sume calls priced at about $0.61 in total.
Cut a YouTube Short on the beat: BPM to cut times in Python
Turn a music bed's BPM into cut start times for a 30-second Short. Runnable Python, plus the $0.10 Timeline render and $0.125 bed.
YouTube Shorts music not working: bake in your own bed
When library music fails on a Short, a bed you generated is part of your file. How to make one for $0.125 and duck it under speech.
Product cutout pop-in for Shorts: RMBG plus compose cost
Cut a product photo out for $0.0225 and overlay it on a clip for $0.02. The per-product price of a pop-in, with the limits.
Batch of 10 product Shorts: cost table on Sume
Ten five-still product Shorts cost $4.675 on Sume, or $6.675 with captions. What scales and what a preflight saves.
Amazon affiliate Short: sales needed to pay back the video
How many sales cover a $0.4675 product Short? A table with assumed commission rates, since Amazon's real rate depends on category.
YouTube Shorts hook generator: what Sume does and does not do
Sume has no hook-generator button. It can test a 3-second opener for $0.1125 on Gemini Omni Flash. Five variants cost $0.5625.
MiniMax H3 native audio vs TTS plus lip sync for a 10-second line
MiniMax H3 makes stereo audio with the clip. For an exact spoken line, Sume prices TTS plus Fabric or H3 Max lip sync: $0.75 to $2.01 for 10 seconds.
AI video with sound: what audio adds per second on Sume
Native audio is a switch on some video models and always on in others. See the per-second price with and without it for Kling 3 Pro, Omni Flash and H3 Max.
Which Sume video models accept an audio reference (audio_url)?
Seedance 2.x, Wan 3.0, MiniMax H3 and H3 Max take audio and video references on Sume. Gemini Omni Flash 1.1 and Recast do not. Rates and limits.
Check your predicted cost against usage.cost when a Sume video ends
Compare your own price prediction with the usage.cost on a completed Sume job and flag drift. A 20-line Python check, tested on three sample polls.
Unit-test a Seedance 2.5 cost estimator against six billed amounts
Your own cost estimate for Seedance 2.5 should match what Sume bills to the micro-dollar. A Python assert test with six known amounts from 480p to 1080p 30 s.
Shorts voiceover: how many characters fill 15, 30, 45 and 60 s
Planning table from script length to seconds and TTS cost, with the 15 characters per second assumption labeled and a way to measure the real value.
YouTube Shorts narrator voice: cost by script length
What a Shorts narrator costs on Sume TTS 1.0 at $0.0475 per 1,000 characters, from a 300-character line to the 20,000-character request cap.
STT 1.0 or video inspect transcribe: two ways to transcribe on Sume
Both bill about $0.01 per audio minute on Sume, but STT takes an audio URL and video inspect takes a Sume clip. Limits of 10 minutes and 1,800 seconds.
EmbeddingGemma 2 audio embeddings or a transcript: what to index
EmbeddingGemma 2 embeds audio directly. Sume has no embedding endpoint, but STT gives text and word timings at $0.01 per minute for search.
YouTube Shorts AI slop: an originality checklist for your pipeline
People search 'youtube shorts ai slop'. A checklist of choices that make a generated Short yours, each tied to a Sume step and its cost.
Keep an AI-use ledger for every Short you upload (Python)
A small CSV ledger that records which Sume steps each Short used and what they cost, so a label decision never relies on memory. Code runs as is.
Submit 40 image jobs with mode webhook in Python on Sume
Send 40 POST /v1/images calls with mode webhook and a webhook_url: each returns a 202 job envelope at once and the result follows. Python, cost, status codes.
POST /v1/images status codes on Sume: 200, 202, 400, 404, 502
What each HTTP status from Sume's image route means and what to do: 200 images, 202 job envelope, 400 unsupported_parameter, 404 model_not_found, 502 failed.
Travel itinerary reel: 4 destinations x 7 seconds on Seedance 2.5
Four 7-second 9:16 destination clips on Seedance 2.5 at 720p cost $4.05 each and $16.20 together on Sume. How to keep one look across all four.
Seedance 2.5 Fast API: Sume has no such id; prices of the four
Searching 'seedance 2.5 fast'? Sume's registry has seedance-2.5, seedance-2, seedance-2-fast and seedance-2-mini. Prices at 480p and 720p, 5 s, 9:16.
8x10 inch print from an AI image: dpi by Sume model at 4:5
An 8x10 print is a 4:5 crop. Pixels and dpi per Sume image model: Seedream 5 Lite 240, GPT Image 2.5 custom 300, Flux 2 Pro 154, with price per image.
Animatic then finals: 12 shots at 480p, 3 hero shots at 1080p
A 12-shot animatic at Seedance 2.5 480p costs $12.96; three 8-second 1080p hero finals add $34.14. Total plan priced on Sume, with the redo caveat.
Which Sume video model is fastest? The catalog's latency labels
Each Sume video model card carries a typical-latency label. Which rows are fast, medium or medium-slow, and what you pay for the faster ones.
Micro-drama episode length: 60, 90 or 180 seconds priced
Reels now allow 3 minutes. See what a 60, 90 and 180-second AI micro-drama episode costs in Wan 3.0 shots and Timeline minutes on Sume.
Seamless looping clip on Wan 3.0: same first and last frame, priced
Wan 3.0 takes a start and an end frame. Use one image for both for a loop: 5 seconds costs $0.32 at 480p, $0.63 at 720p, $1.25 at 1080p.
Add captions to a silent clip: cues instead of speech recognition
A clip with no speech fails caption_no_speech. Send cues with your own text and times to burn on-screen lines on silent or music-only video for $0.20.
Seedance 2 mini vs Wan 3.0: price per second at 480p and 720p
Wan 3.0 is $0.0625 a second at 480p and $0.125 at 720p. Seedance 2 mini works out near $0.088 and $0.189. Ten-second and 15-second costs side by side.
Mistral Large 4 preview: Sume's agent model field takes one value
Mistral Large 4 is reported in preview. Sume lists no Mistral Large id; Agent Completions accepts only sume-agent. What to send instead, and what it bills.
EmbeddingGemma 2 for video search: Sume has no embeddings endpoint
EmbeddingGemma 2 is reported as a 740M, Apache 2.0 multimodal embedder. Sume sells no embeddings API as of 2026-10-07; here is what it does sell for search.
ChatGPT Image 2.5 xhigh and max price grid, six sizes on Sume
What xhigh and max cost next to high on openai/gpt-image-2.5, from 1024 square to 4K: $0.117 and $0.264 at 1024x1024, $0.50 for max at 4K, with the default.
Korean captions: language ko does not choose the style
Setting language to ko on a caption job does not switch the style. The default is black-outline; a Latin style rejects Hangul with a 400. What to send instead.
How fresh are Sume's video prices? Rate-card dates per model
Every Sume video price traces to a dated provider rate card from June to September 2026. The date per model, and what 'subject to change' means for a budget.
90-second micro-drama episode: six H3 shots or three Wan shots, priced
Micro-dramas keep rising on TikTok, Reels and Shorts. A 90-second episode costs $6.98 on six H3 clips or $11.45 on three Wan 720p clips.
Go MaxBytesReader: cap a Sume webhook body at 1 MiB, 413 before HMAC
A webhook endpoint that reads an unbounded body invites memory abuse. A Go handler with http.MaxBytesReader, an HMAC check and a refusal of empty secrets.
Listing walkthrough from 8 photos: seven 4-second Wan 3.0 transitions
Eight room photos become seven Wan 3.0 first-to-last-frame transitions of 4 seconds. $3.50 at 720p plus $0.10 for the timeline render on Sume.
Swift withTaskGroup: four Sume video submits in flight
A sliding window in Swift 6: withTaskGroup keeps four POST /v1/videos requests in flight, re-arming as each finishes. One file, URLSession, no packages.
Ruby Thread and Queue: four workers submit Wan 3.0 clips
Ruby worker pool for POST /v1/videos: a Queue of shots, four Threads, Net::HTTP and one Idempotency-Key per shot. Tested on Ruby with a stub server.