Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
Loop a 30-second music bed under a 3-minute Reel: the body
A short bed can cover a 180-second Reel with soundtrack.loop, duck_db and a fade-out. One Timeline 1.0 body, the 3-minute bill, and the limits to respect.
LTX-2.5 audio-to-video: three ways to drive a clip with audio on Sume
LTX-2.5 is not on Sume. To make video that follows a voice or track, use audio references on Seedance 2.5, Wan 3.0 or H3, or H3 Max lip sync. Limits and prices.
LTX-2.5 or a lip-sync API for a talking head: what Sume offers
Sume does not list LTX-2.5. For a talking head that must say your exact words, the route that ships is TTS plus H3 Max lip sync, or Avatar 1.0 talking video.
MAI-Voice-2.1 is not on Sume: make talking clips with Sume TTS
Microsoft MAI-Voice-2.1 launched 2026-10-01 but Sume does not list it. Here is the path that ships: Sume TTS audio, then H3 Max lip sync on a still.
MAI-Voice-2.1 or Sume TTS? Three questions that decide it
Live speech, extra outputs or lowest price per character? A short guide to MAI-Voice-2.1 and Flash against the Sume TTS Router, with 10M-character math.
Make an AI voiceover louder: generation_config volume and a peak check
Sume TTS takes generation_config.volume from 0.5 to 2. Raise it, then check the WAV peak in Python so you never ship a clipped voiceover.
MiniMax H3 lip sync at 2K? Sume offers 480p, 768p and 1080p
MiniMax H3 the video model lists 2K, but Sume's H3 Max lip-sync route stops at 1080p. The three resolutions, their derived per-second prices and how to choose.
MiniMax H3 Max video or H3 Max lip sync: fixing the wrong_tool error
generate_video refuses minimax/h3-max/lip-sync with wrong_tool. Use avatar-image-to-video_create for lip sync, minimax-h3-max for text or image video.
MiniMax H3 open weights vs a hosted lip-sync API: what you take on
MiniMax released H3 open weights on 2026-08-03. Self-hosting is not the same as a hosted still-plus-audio lip-sync route. What each choice makes you own.
MiniMax H3 reference limits: 9 images, 3 videos, 3 audios on Sume
MiniMax H3 reference-to-video on Sume takes up to 9 images, 3 videos and 3 audio clips, 12 in total. The duration rules, the extra-image fee and a request.
Native 4K AI video API: LTX-2.5 claims 4K, what Sume offers instead
LTX-2.5 is listed up to 4K but is not in Sume's catalog. Sume's 4K options are Gemini Omni Flash 1.1 and MiniMax H3 4K upscales. Limits and per-second prices.
AI video with native audio vs a scripted voiceover: exact words
Seedance 2.5, MiniMax H3 and Kling make sound with the picture. When a line must be word-exact, generate the voiceover with Sume TTS and join it on a timeline.
Nearest supported aspect ratio per Sume image model, in Python
A model rejects an aspect ratio it does not list. Read the catalog, pick the closest ratio it accepts, generate, then crop to the exact shape you need.
One API key for Seedance 2.5, Wan 3.0, Kling 3 and MiniMax H3
Do you need a ByteDance, Alibaba, Kuaishou and MiniMax account to call their video models? On Sume one key and one wallet cover all four. How it works.
One 6-minute b-roll, twelve Shorts episodes: source_in offsets
Slice one imported b-roll into twelve non-repeating 30-second episode backgrounds with Timeline 1.0 source_in, and plan the whole season unbilled in Python.
opencode remote MCP entry for Sume: env syntax and a longer timeout
The hosted Sume server in opencode.json: type remote, a bearer header from an env variable, a timeout above jobs_wait's 55 s cap. Checked by a script.
Guardrails for an agent calling paid APIs: key, cap, and what to log
An agent calling a paid media API needs three habits: an idempotency key on each write, a spend cap per run, and logs that hold ids, never signed URLs or keys.
Photo to watercolor or pencil sketch with an API, composition kept
Turn a photo into a watercolor or pencil drawing on Sume: send it as an input reference to GPT Image 2.5 and say what to keep. There is no strength slider.
Pinterest visual search ads: how many image variants per call?
Sume's image API takes n up to 10 per call, with lower per-model ceilings. Read the n range from the endpoint record before you plan a variant set.
Pinterest visual search ads: a transparent-background product cutout
Visual search ads match your product by its look. Ask for background: transparent on openai/gpt-image-2.5 and keep a clean PNG cutout of each item.
Python exceptions for Sume errors: retry on the class, not the status
Map the Sume error envelope code to two exception classes, Retryable and Fatal, so one except clause drives retries. Covers 409, 429, 402 and 503.
Python: run one prompt on four AI video models and save the clips
A Python script that sends one prompt to Wan 3.0, Seedance 2.5, Kling 3 and MiniMax H3 on Sume, polls all four jobs and saves each MP4, with costs.
Read Sume TTS raw PCM in Python: f32le samples at 24 kHz
Ask the TTS Router for container raw, 24000 Hz and pcm_f32le, then load the samples with the standard library for DSP or a custom player.
Reddit 15-second ad: join hook, demo and offer, then plan first
Build a 15-second ad from three clips on Timeline 1.0 and check its length and cost with the unbilled /plan call before you pay for a render.
Reddit 15-second engaged views: does your voiceover CTA land?
Reddit's Engaged Video Views beta bills videos over 15 s at 15 s. Use TTS word timestamps to check that your call to action is spoken before second 15.
Reddit's 15-second view rule: a lip-sync clip that stays under it
Reddit's Engaged Video Views beta bills videos over 15 s at 15 s. Sume H3 Max lip sync tops out at 14.8 s of audio, so one clip fits. Specs, math and a caveat.
Reddit's 15-second video views: what a 15 s clip costs on Sume models
Reddit bills videos over 15 s at 15 s in its Engaged Video Views beta. A 15-second clip costs $1.88 on Wan 3.0 at 720p or $2.10 on Kling 3 without sound.
Reel captions in two languages: language hint vs auto-detect
How Sume video captions pick a language and a style: the optional language hint, auto-detect, the Hangul rule that returns 400, and a Korean Reel example.
Render a Short from an agent: hosted MCP tool order
Which Sume hosted MCP tools an agent calls, in order, to import, inspect, cut and render a vertical Short, with idempotency_key rules and what stays on REST.
Menu refresh, 30 dishes: a dish card and a 5-second clip each on Sume
A seasonal menu of 30 dishes with one Ideogram 4.5 dish card and one 5-second Wan 3.0 clip each: $11.85 at 480p, $21.15 at 720p, plus one $0.125 music bed.
Restaurant table tent art: Ideogram 4.5 at 3:4, specials quoted
Make table tent artwork for a restaurant: a 3:4 Ideogram 4.5 image with each special quoted in the prompt, a price check, and the real cost per tent on Sume.
Retry only the failed episodes from a Sume bulk queue
A Sume bulk queue shows completed even when items failed. Find the failed children and re-queue only those, with fresh keys and untouched keys for the rest.
Sale badge on a product image: draw it in Pillow, not in the prompt
Generate a clean product photo on Sume, then draw the sale badge and percentage from your catalog data in Pillow, so a price change never costs a regeneration.
Save a Sume artifact atomically: write a .part file, then rename
A half-written MP4 that looks finished is worse than none. Download a Sume artifact to a .part file, check the length, rename once. Tested in Python.
Season 2 of a Shorts series: same Sume Format, new input, new keys
Start season 2 without rebuilding anything. Keep the Format, change the input and idempotency keys, and keep season 1 receipts as a style reference.
Seedance 2.5 job failed: refund, new idempotency key, and rerun cost
A failed Seedance 2.5 job is refunded on Sume. Retry with a new Idempotency-Key; the old one replays the failed job. A Python handler and the rerun cost.
Low-latency TTS without streaming: one job per sentence
Sume TTS has no streaming. To start playback early, split the script by sentence, submit the jobs in parallel and play each file as it finishes, in order.
Shortest and longest values Sume accepts for a Short: one table
Minimum clip, slot, spine and fade values and the maximums for trim, transitions and soundtrack on Sume, in one table with an offline slot checker in Python.
Shorts episode title card: a 2-second still, then the clip
Open each Shorts episode on a held title still with Timeline 1.0: slot math, the 0.5 s coverage rule, what stills ignore, and one render body you can copy.
Shorts season webhooks: a Python receiver that counts episodes
Receive Sume job webhooks for a season of Shorts renders: verify sume-v1, reject an empty secret, de-dupe by job_id, and know when every episode is terminal.
Shorts season trailer: first 5 seconds of every episode in one clip
Build a Shorts season trailer by trimming 0 to 5 s from each episode with video trim, then planning one timeline render under a music bed. Python included.
Shorts series bulk queue finishes out of order: publish by number
Episodes in a Sume bulk queue run in parallel and finish in any order. Sort by an episode number in your output schema, not by completion time.
Shorts spec sheet against Sume Timeline output: what is confirmed
YouTube lists 3 minutes, 1080p and a 100-character title for Shorts. Sume Timeline renders 1080x1920 by default. A table of what is confirmed and what to check.
Sora API gone: 9:16 portrait video on Sume, model by model
Which Sume video models return 9:16 after the Sora API ended on 2026-09-24, how to ask for it, and a script that reads the catalog instead of trusting a table.
Sora API ended: which Sume route replaces a talking presenter clip?
OpenAI's Sora API ended 2026-09-24. For a presenter who says exact words, the replacement is a still-plus-audio or avatar route, not a video model.
Failed AI video job on Sume: is the reservation released?
Sume reserves the estimated cost at submit and captures it on success; failed jobs release or refund it where applicable. How to check your own account.
GitHub Actions job that renders a 9:16 clip on Sume as an artifact
A workflow file that replaces a Sora render step: submit to Sume, poll with a deadline, and upload the mp4 as a build artifact. The key stays in a secret.
Seedance 2, Omni or Wan 3.0 for 9:16 ads on Sume: how to choose
Three Sume video models side by side for vertical ads after Sora: clip length, resolutions, 9:16 in the docs, references, and the price of a 5 s clip.