Guides
Sume guides: the API, Formats, agents, models, media tools and pricing, each answer dated and sourced.
Ideogram 4.5 vs Nano Banana 2 for multi-turn edits: cost per turn
Ideogram 4.5 claims clean multi-turn edits. Compare it with Nano Banana 2 on Sume: price per turn, six-turn totals and a drift test you can run.
Kajabi lesson video from a 16:9 Sume avatar clip: what fits
Kajabi takes MP4 up to 4 GB and recommends 16:9. A 16:9 Sume avatar clip of up to 60 seconds is a lesson intro or recap. The numbers, and where it stops.
Laravel queued job for an AI video API: delay, redispatch, poll
A Laravel ShouldQueue job reads a Sume video job once and redispatches itself with ->delay() from next_poll_after_seconds, so no worker sleeps during a render.
Let QA override the image model per request, with an allowlist
An internal render endpoint that honors an X-Image-Model header only for allowlisted Sume ids, so QA can test gpt-image-2.5 before the config flips. Python.
Monthly AI content budget at three sizes: starter, growth and scale
Three monthly volumes of images, voiceovers and 6-second clips priced on Sume: the starter plan is about $10, growth about $39 and scale about $146.
Nova Canvas cfgScale 1.1 to 10 vs Sume: no guidance field
Nova Canvas has a cfgScale from 1.1 to 10, default 6.5. Sume rejects fields a model does not list. How to get the same effect with wording and model choice.
Nova Canvas inpainting mask rules vs a Sume reference edit
Nova Canvas inpainting wants a pure black and white mask, same size as the input. What to build, and how Sume's edit call differs.
Nova Canvas returned fewer images than asked: Sume n and data length
Nova Canvas can return fewer images than numberOfImages when moderation blocks some. How to count what you got and what you paid for on Sume.
Nova Canvas rejects PNGs with transparent pixels: flatten before edits
Nova Canvas needs 8-bit PNG or JPEG with no transparent pixels in the alpha channel. Flatten cutouts onto white before any image edit, here and on Sume.
Nuxt Nitro server route for an AI video webhook with readRawBody
A Nuxt server/api route calls readRawBody, checks Sume's sume-v1 HMAC and answers 401 on a mismatch before it queues the finished video download.
One Sume TTS job: 20,000 characters is about 26 minutes of narration
At 750 characters per minute, a 20,000-character job covers about 26.7 minutes for 95 cents. Compare Shorts at 3 min, TikTok at 10 min and Reels at 20 min.
Outbrain outstream video ad: 60 s, 250 MB, a Sume avatar clip
Outbrain outstream video takes up to 60 seconds and 250 MB in 9:16, 1:1 or 16:9. A Sume avatar job fits the length and the ratios. Check the file before upload.
Output schema unsupported_ref: $defs must live at the schema root
A $ref to a nested $defs, an external URL or a missing name fails with unsupported_ref in a Sume output_schema. Move $defs to the root and fix the pointer.
Output schema max_depth: 10 levels, and $defs recursion is free
Sume's output_schema allows 10 levels of literal nesting. Past that you get max_depth. A self-referencing $defs entry does not add to the depth count.
Output schema node with only a description: missing_type violation
A property with only a description is legal JSON Schema, but Sume rejects it with missing_type. Give it a type, a $ref or an anyOf, and move constraints.
Output schema type float or date rejected: unsupported_type
Sume's output_schema accepts seven types only. A float, date or any type gets the rule unsupported_type. Map each to number, string with format, or an enum.
Podcast network: transcripts, cover art and stings for 12 episodes
12 episodes of 45 minutes, four Ideogram 4.5 cover options each and three music stings each cost about $14.24 a month on Sume. Unit prices and the sum.
Print-on-demand: 200 designs a month on Ideogram 4.5, plus cutouts
200 text-on-shirt designs a month cost about $16 at Ideogram 4.5 medium or $58 at high on Sume, plus $4.75 for cutouts. The budget, line by line.
Monthly AI media budget calculator in Python, from live Sume prices
Build a monthly budget for images and voiceovers by reading list prices from the Sume image and TTS router catalogs, then applying list x 1.25 and the 5.5% fee.
Python TTS cost calculator: Sume job rounding vs per-character rates
A runnable Python function that prices narration lines on Sume (cent rounding, 1-cent minimum) and at flat per-million rates for MAI-Voice-2.1 and Flash.
Reels profile grid crops to 3:4 (1080x1440): check a frame
A Reel cover is center-cropped to 3:4 on the profile grid, per a third-party guide. Pull the frame with Sume video frames and keep faces and titles inside.
Retired AI model id to Sume id: GPT Image, Gemini and Sora lookup
One lookup table that maps retired gpt-image-1, gemini-2.5-flash-image and sora-2 ids to Sume ids, with dates, so a failing call tells you what to switch to.
Schedule the next Ideogram 4.5 batch wave from ratelimit-reset
Size each wave from ratelimit-remaining and wave_size_hint, and when the write bucket is empty sleep ratelimit-reset seconds. A pure function you can test.
How many characters is a 15, 30, 45 or 60 second AI narration
Using 750 characters per minute of speech, a 15-second voiceover is 188 characters, 30 seconds is 375, 60 seconds is 750. Sume TTS cost for each.
Start a song at the chorus in a Short: split, then soundtrack
Timeline's soundtrack has no in-point. Cut the chorus with Timeline audio split, then pass the new audio_url as the soundtrack. Steps and cost: $0.01 + $0.10.
Sume 401 halfway through a batch: stop every worker, do not retry
A 401 on a Sume submit means a missing, malformed or revoked key. The SDK does not retry it, and neither should you. Python sample that keeps job ids.
Sume schedule slug rules: 2-64 characters, and runs is reserved
A Sume schedule slug is lowercase alphanumerics with single hyphens, 2 to 64 characters, unique in your account. The word runs is reserved. See the vanity path.
Sume SDK maxRetries: 0 when your job queue already retries the call
The SDK retries 408, 429 and 5xx twice by default. Under a queue with five attempts that is up to 15 tries. Set maxRetries to 0 and let one layer retry.
Sume waitForJob pollInterval is a floor; next_poll_after_seconds wins
waitForJob never polls faster than pollInterval, and a longer next_poll_after_seconds from the server raises the gap. Defaults, timing table, sample.
Sume webhook handler over 10 seconds: acknowledge first, then work
Sume gives each webhook attempt 10 seconds. Verify, answer 2xx, then process in the background and dedupe on job_id. Node sample you can run locally.
Swap the AI image model without a redeploy: JSON config hot reload
Read the Sume image model id from a JSON file that reloads when it changes, so a gpt-image-1 shutdown fix is a one-line edit with no deploy. Python, stdlib.
TTS API price per 1,000 characters: Sume, MAI, Groq, Lemonfox, Unreal
A per-1,000-character table of list prices for Sume, MAI-Voice-2.1 and Flash, Groq Orpheus, Lemonfox and Unreal Speech, with cost for a 750-character short.
Sume Timeline output: H.264, AAC 192k and 1-second keyframes
What a Timeline 1.0 MP4 contains for a Reel, Short or TikTok upload: libx264 CRF 20, yuv420p, AAC 192k, faststart, 1-second keyframes, and what you cannot set.
Timeline segment_source_short_looped vs padded: loop or freeze a clip
When a clip is shorter than its slot, Sume Timeline either loops it or holds its last frame. How render.pad_mode auto, loop and freeze decide, with thresholds.
Titan base64 image fields vs Sume input_references URLs in Python
Titan sends images as base64 strings inside the request. Sume wants public HTTPS URLs. A Python adapter from a local file to a URL for edits.
Titan Image Generator v2: 1,408 px and 5 MB limits vs Sume edits
Titan Image Generator v2 caps edit inputs at 1,408 by 1,408 px and 5 MB. How to resize before Bedrock, and what a Sume edit call needs instead.
Make a Top 5 vertical video from five stills with Sume Timeline
Five images, a 30-second silent Timeline at 1080x1920, then burned captions: the slot math, the still-motion limit and the cost in Sume.
tts_sentence_selection_invalid 422 on Sume TTS: what triggers it
Sume TTS returns 422 tts_sentence_selection_invalid for gaps, repeated jobs, unfinished jobs and partial coverage. Each cause and its fix.
tts_source_not_found 404 on Sume TTS: revision, sentence or job
A 404 tts_source_not_found from the Sume script-source API means the revision, a sentence id or a selected job is not visible to this key or thread.
tts_source_too_large 422: Sume TTS 20,000-character cap on a selection
A transcript_source selection that resolves past 20,000 characters returns 422 tts_source_too_large. Split it by sentence ids and price each job.
4K vertical Short in Timeline: the 2160 cap and 1214x2160
Timeline output width and height top out at 2160 and must be even, so 2160x3840 is refused. What the largest 9:16 frame is and whether a Short needs it.
waitForJob on a failed Sume image job: read the record, skip /result
waitForJob resolves for failed and canceled jobs instead of throwing, and /result returns 409 job_not_completed for them. How to branch on job.status.
Sume TTS source errors: which are safe to retry (status table)
Every tts_ error code in Sume's script-source API with its HTTP status, whether it charges, and whether a retry can help. A table for client error handling.
Seedance 2.5 vs 2.0 vs Fast vs Mini: price per clip at 720p on Sume
At 720p a 10-second clip costs $5.78 on Seedance 2.5, $3.78 on Seedance 2, $3.03 on Fast and $1.89 on Mini on Sume. Only 2.5 reaches 30 seconds.
Is ultrawide 21:9 pricier on Seedance 2.5 than 16:9 on Sume?
On Sume, Seedance 2.5 at 21:9 or 4:3 costs a cent or a few more than 16:9, 9:16 or 1:1. A 30-second 720p clip is $17.42 at 21:9 against $17.34.
How much does a 30-second Seedance 2.5 clip cost on Sume?
A 30-second Seedance 2.5 clip costs $8.07 at 480p, $17.34 at 720p and $42.65 at 1080p on Sume. Prices from Sume's pricing code with the 1.25 margin.
How much does a 40-second AI video cost on Sume, by model and split?
A 40-second sequence costs $2.51 on Wan 3.0 at 480p, $5.00 at 720p and $23.12 for Seedance 2.5 at 720p. Prices for each split into legal-length jobs.
Does native audio cost extra on Wan 3.0 and Gemini Omni Flash on Sume?
On Sume, Wan 3.0 costs the same with the audio flag on or off, and Gemini Omni Flash 1.1 always includes audio. Billed prices for both, from Sume's code.